Spin Off: AT's Testing Methods & Uber Mode

[Keysplayr]

n/m

 

[raghu78]

the point is the noise issue was overblown by ryan. i am sure he got the message by now. other reviewers like scott of techreport and brett of hardocp said the ref cooler was not too loud at uber and both said they spent considerable duration of time gaming on the R9 290/ R9 290X before forming their conclusions or opinions. remember both acknowledged the ref cooler could have been better. but they also acknowledged the outstanding value proposition and said all things considered the R9 290 was an excellent card at USD 400. this is a balanced take.

finally noise is both an objective and subjective topic. what is acceptable noise can differ from person to person. ryan cannot talk for his entire audience. maybe post a video of the fan under load. he should leave the objective data for user interpretation. ryan is there to review a product and objectively point out its pros and cons and let the user decide if he likes it or not and wants to buy it or not. he is not there to recommend anything. period.

 

[KingFatty]

I think fair minded people with intelligence understand why Ryan did what he did and does reviews the way he does and understand that fault is not with Ryan or AnandTech and it's policies....but with AMD for not launching the polished product they should have.

Have you read the follow up? I was impressed how it seems to address many of the issues raised here, and it seems like there was very good feedback communication. So maybe that can be seen as some kind of acknowledgement that there was some merit to this discussion?

So, I appreciate that someone might see this a whiny complaining, but apparently the feedback was heard and made a difference.

However, there is some careful word twisting that further soured me:

In the original article, it says:

To get right to the point then, this is one of a handful of cards we’ve ever had to recommend against.

Notice how he is actively against the card, going beyond a mere failure to recommend.

But in the follow-up, he re-characterizes his words by saying:

In this week’s article I flat out avoided recommending the 290 because of its acoustic profile.

See the change, like it's an attempt to re-write the negativity? A failure to recommend is different than actively recommending against.

Maybe the original words were so harsh, it's as though the memory is shocked into denying their reality? Just kidding, I'm sure there is some ego protection going on, and perhaps the writing was a bit coerced/requested by higher-ups and not entirely spontaneous?

 

[VulgarDisplay]

The words of a reviewer that is unbiased would never be to "recommend against." An unbiased reviewer in the case of the 290x/290 would simply state, "I can only recommend buying this card if you are ok with the additional noise over the gtx780/780ti."

Especially with such a subjective metric to be so against. If AMD was selling a card that offered 1fps in all games for $2000 dollars then by all means say "recommend against" because it's a no brainer to say it then. When the GPU offers Titan level performance for $400 with the caveat of a little extra noise just tell them about the noise. Leave out the bombastic personal opinion.

My advice to Ryan is to be careful in his upcoming reviews. People on both sides are going to be looking for any little slip in his writing moving forward. Not a good path to be going down for a reviewer who depends on his reputation to stay employed.

 

[raghu78]

The words of a reviewer that is unbiased would never be to "recommend against." An unbiased reviewer in the case of the 290x/290 would simply state, "I can only recommend buying this card if you are ok with the additional noise over the gtx780/780ti."

Especially with such a subjective metric to be so against. If AMD was selling a card that offered 1fps in all games for $2000 dollars then by all means say "recommend against" because it's a no brainer to say it then. When the GPU offers Titan level performance for $400 with the caveat of a little extra noise just tell them about the noise. Leave out the bombastic personal opinion.

My advice to Ryan is to be careful in his upcoming reviews. People on both sides are going to be looking for any little slip in his writing moving forward. Not a good path to be going down for a reviewer who depends on his reputation to stay employed.

well said :thumbsup: ryan flat out accepted he should not have spoken for himself rather left the audience to ponder on the tradeoffs.

http://www.anandtech.com/show/7501/amd-changes-290-series-fan-algorithms

"I personally value acoustics very highly and stand by my original position that the reference R9 290 is too loud. When I game I use open back headphones so I can listen for phone calls or the door for shipments, and as a result acoustics do matter to me. In the review I assumed everyone else valued acoustics at least similarly to me, but based on your reaction it looks like I was mistaken. While a good number of AnandTech readers agreed the R9 290 was too loud, an equally important section of the audience felt that the performance delivered was more than enough to offset the loud cooling solution. We want our conclusions to not only be reflective of our own data, but also be useful to all segments of our audience. In the case of the 290 review, I believe we accomplished the former but let some of you down with the latter."

the worst part is even the R9 290 series card owners ( and i have heard from quite a few on ocn) are comfortable with 55% uber fan speed as scott of techreport and brent of hardocp mentioned. this goes to further prove how much of a mistake it was for Ryan to think everyone shares his same idea of whats acceptable noise levels for a GPU.

http://www.overclock.net/t/1440501/to-buy-r9-or-finnish-my-sli-plan/10#post_21158879

 

[railven]

the point is the noise issue was overblown by ryan. i am sure he got the message by now. other reviewers like scott of techreport and brett of hardocp said the ref cooler was not too loud at uber and both said they spent considerable duration of time gaming on the R9 290/ R9 290X before forming their conclusions or opinions. remember both acknowledged the ref cooler could have been better. but they also acknowledged the outstanding value proposition and said all things considered the R9 290 was an excellent card at USD 400. this is a balanced take.

finally noise is both an objective and subjective topic. what is acceptable noise can differ from person to person. ryan cannot talk for his entire audience. maybe post a video of the fan under load. he should leave the objective data for user interpretation. ryan is there to review a product and objectively point out its pros and cons and let the user decide if he likes it or not and wants to buy it or not. he is not there to recommend anything. period.

The words of a reviewer that is unbiased would never be to "recommend against." An unbiased reviewer in the case of the 290x/290 would simply state, "I can only recommend buying this card if you are ok with the additional noise over the gtx780/780ti."

Especially with such a subjective metric to be so against. If AMD was selling a card that offered 1fps in all games for $2000 dollars then by all means say "recommend against" because it's a no brainer to say it then. When the GPU offers Titan level performance for $400 with the caveat of a little extra noise just tell them about the noise. Leave out the bombastic personal opinion.

My advice to Ryan is to be careful in his upcoming reviews. People on both sides are going to be looking for any little slip in his writing moving forward. Not a good path to be going down for a reviewer who depends on his reputation to stay employed.

The hell are you guys talking about? When a reviewer reviews anything they give it an aggregated score based on whatever system they have in place.

Their conclusion is ALWAYS a recommendation "this movie is a must see", "this game is an absolute must own", "avoid this car at all costs."

I've never read a review where a recommendation isn't made, what do you think those stupid badges websites like to use now like "Golden Editor's Choice Award" you think that's an objective image?

Ryan is not wrong in saying what he said, the card was louder than his personal threshold hold could tolerate, he didn't recommend the card on this alone, but the data is there for you to interpret and decide whether you agree with Ryan.

However, what Ryan did with the 780 Ti review is basically throw out a whole subset of data he collected under the guise of "it isn't out of the box performance."

Don't confuse the two things, notice this thread is about UBER MODE, the 290 doesn't have Uber mode.

 

[blackened23]

the point is the noise issue was overblown by ryan.

Did I seriously just read that? I can't read this with a straight face, that the AMD noise issues are overblown. At 40%, yes, it is quiet. At 55%? If you think 55% or anything past 45% is quiet on an AMD reference shroud, you're basically being disingenuous. I've owned 3 generations of ATI cards and I know what they sound like, and they're all identical under the hood - even the 290X shroud is the SAME as the 5870, 6970, and 7970 shroud. IT IS NOT quiet past 45%, it is especially not quiet compared to any Kepler shroud, and it is VERY VERY stupid that to get a quiet card that you must lose 20% performance.

The fact that you lose 20% of your performance JUST FOR a quiet fan profile is so absurd that I cannot believe people aren't more critical about AMD. There has never, in the history of GPUs, been a card that shed 20% performance for a quiet fan speed. I feel like half the people arguing in favor of the 290 shroud probably haven't owned any AMD GPU because anyone that has, knows what it sounds like, and that sound ISNT quiet past 45%....

Now with my 7970s and prior ATI cards, I could overlook this. In fact, I freaking loved my 7970s in the brief time I had them. Do you know why? I could overclock the bejebus out of my 7970s past 1100mhz without breaking a sweat and without using a stupid high fan speed. My cards were still quiet EVEN WHEN overclocked, I used 40-43% fan speed IIRC. It was acceptable in terms of acoustics - it wasn't Kepler quiet, but it was very good. Now with the 290 series of cards, you basically have to use a 55-65% fan cap (per BrentJ) to prevent any sort of throttling. I'm sorry, but that is BS. Pure BS. That's what gets me the most - I could overlook the 290 using this shroud MAYBE but with this throttling nonsense? I don't think so. AMD needs to be criticized about this because the Hawaii chip is great yet they failed to deliver on everything else about the card. AMD needs to get this message that this type of garbage, the throttling vs noise compromise is not acceptable. Oh yeah, before you say it, passing the buck to AIB makers isn't the answer either because not everyone can use an open air cooler.

I truly hope AMD gets this message somehow - they need to know that user experience matters, even if the price of the card is 20$ higher. As things are, Hawaii is an *amazing* chip but it is so marred by that reference shroud that it blows my mind. 20% performance gone for the quiet fan profile. Like I said, that is BS. This card had the potential to be THE CARD OF ALL CARDS and that opportunity was completely thrown out the window - like I said, great chip, terrible user experience especially with the quiet mode throttling.

 

[VulgarDisplay]

The hell are you guys talking about? When a reviewer reviews anything they give it an aggregated score based on whatever system they have in place.

Their conclusion is ALWAYS a recommendation "this movie is a must see", "this game is an absolute must own", "avoid this car at all costs."

I've never read a review where a recommendation isn't made, what do you think those stupid badges websites like to use now like "Golden Editor's Choice Award" you think that's an objective image?

Ryan is not wrong in saying what he said, the card was louder than his personal threshold hold could tolerate, he didn't recommend the card on this alone, but the data is there for you to interpret and decide whether you agree with Ryan.

However, what Ryan did with the 780 Ti review is basically throw out a whole subset of data he collected under the guise of "it isn't out of the box performance."

Don't confuse the two things, notice this thread is about UBER MODE, the 290 doesn't have Uber mode.

Re read my post. I didn't say a reviewer can't give a recommendation. I said an unbiased review should be very careful in how to recommend, or recommend against a product. The language used in the reviews was not carefully worded. Frankly, constant focus of the reviews on noise was not necessary.

Everyone agrees the cooler could be better which would in turn make the GPU more quiet. People came to that conclusion after reading countless other reviews that did not smash the reader over the head with a "NOISE BAD" hammer like Ryan's review at Anandtech.

 

[Will Robinson]

blackened23
Diamond Member



Join Date: Jul 2011
Posts: 6,641

Quote:
Originally Posted by raghu78
the point is the noise issue was overblown by ryan.

Did I seriously just read that? I can't read this with a straight face, that the AMD noise issues are overblown. At 40%, yes, it is quiet. At 55%? If you think 55% or anything past 45% is quiet on an AMD reference shroud, you're basically being disingenuous. I've owned 3 generations of ATI cards and I know what they sound like, and they're all identical under the hood - even the 290X shroud is the SAME as the 5870, 6970, and 7970 shroud. IT IS NOT quiet past 45%, it is especially not quiet compared to any Kepler shroud, and it is VERY VERY stupid that to get a quiet card that you must lose 20% performance.

The fact that you lose 20% of your performance JUST FOR a quiet fan profile is so absurd that I cannot believe people aren't more critical about AMD. There has never, in the history of GPUs, been a card that shed 20% performance for a quiet fan speed. I feel like half the people arguing in favor of the 290 shroud probably haven't owned any AMD GPU because anyone that has, knows what it sounds like, and that sound ISNT quiet past 45%....

Now with my 7970s and prior ATI cards, I could overlook this. In fact, I freaking loved my 7970s in the brief time I had them. Do you know why? I could overclock the bejebus out of my 7970s past 1100mhz without breaking a sweat and without using a stupid high fan speed. My cards were still quiet EVEN WHEN overclocked, I used 40-43% fan speed IIRC. It was acceptable in terms of acoustics - it wasn't Kepler quiet, but it was very good. Now with the 290 series of cards, you basically have to use a 55-65% fan cap (per BrentJ) to prevent any sort of throttling. I'm sorry, but that is BS. Pure BS. That's what gets me the most - I could overlook the 290 using this shroud MAYBE but with this throttling nonsense? I don't think so. AMD needs to be criticized about this because the Hawaii chip is great yet they failed to deliver on everything else about the card. AMD needs to get this message that this type of garbage, the throttling vs noise compromise is not acceptable. Oh yeah, before you say it, passing the buck to AIB makers isn't the answer either because not everyone can use an open air cooler.

I truly hope AMD gets this message somehow - they need to know that user experience matters, even if the price of the card is 20$ higher. As things are, Hawaii is an *amazing* chip but it is so marred by that reference shroud that it blows my mind. 20% performance gone for the quiet fan profile. Like I said, that is BS. This card had the potential to be THE CARD OF ALL CARDS and that opportunity was completely thrown out the window - like I said, great chip, terrible user experience especially with the quiet mode throttling.

Look,everyone understands you have just spent a lot of money on that 780SC.
Time marches on,faster,cheaper cards come out.
Seeing you mentioned "The entire history of GPUS!!!! I'm sure you've
been around long enough to understand that.:thumbsup:

I truly hope AMD gets this message somehow - they need to know that user experience matters, even if the price of the card is 20$ higher.

I'm pretty sure no one will have missed your message(s)....

 

[Janooo]

I hate to break this to you but R9 290($400) owns all of NV cards from 780 down and R290X owns the NV flagship Titan as well.(remember the $1000 price tag Titan?)
780Ti barely scrapes past 290X and is SEVEN HUNDRED DOLLARS.
It's not AMD fans who are "angry"

What are you talking about?
Check the post #145.
The 290X in Uber mode is faster than Ti based on AT's own numbers.
That's the reason why AT chose to ignore the Uber mode and the only way AT can tell Ti is faster card.
Ryan says "our policy is to always compare out of the box performance when possible" just a few days after he invented "Noise Equalization" at 34% fan speed that does not come out of the box.

That's a dishonest manipulation with AT's readers.
AT should apologize for this!

 

[blackened23]

What? All reviews have an element of subjectivity. If they didn't, every movie review would be a 100%. Every restaurant review would be glowingly positive.

In the real world, if a reviewer finds something questionable about a product the review will make note of it. Period. It sounds like you're upset that the 290 didn't get a gold award or something along those lines. AMD needs to be hammered over their 290 series of GPUs throttling by nearly 300mhz while a quiet mode profile is in use - that is so absurd it blows my mind. I did a double take when I saw the Kepler throttle by like 13-26mhz back in 2012 - and then when I figured the GPU boost system out, it made sense. 13-26MHz doesn't affect performance to a perceptible extent and merely keeps efficiency in check. But here we are with AMD throttling by nearly 300mhz for a quiet fan profile. And you guys are acting like, it's just cool. 300mhz. It's cool guys, 20% performance down the drain! Where's the gold award for the 290, Ryan?!

 

[Subyman]

The fact that you lose 20% of your performance JUST FOR a quiet fan profile is so absurd that I cannot believe people aren't more critical about AMD. There has never, in the history of GPUs, been a card that shed 20% performance for a quiet fan speed. I feel like half the people arguing in favor of the 290 shroud probably haven't owned any AMD GPU because anyone that has, knows what it sounds like, and that sound ISNT quiet past 45%....

Your comment actually doesn't make any sense. Of course you have never seen a GPU drop performance using a quiet mode, because there has never been a quiet mode option on a reference card before...

 

[blackened23]

Your comment actually doesn't make any sense. Of course you have never seen a GPU drop performance using a quiet mode, because there has never been a quiet mode option on a reference card before...

I'm talking manual fan speeds. You know, there have been utilities for adjusting manual fan speed to comfortable levels, for YEARS.............and there hasn't been a GPU that tanked your performance for using a reasonably quiet fan speed. Not the 5870, not the 6970, not the GTX 280, etc......

Only you cannot use a quiet fan speed on the 290. Unless you want to say goodbye to 20% of your performance.

 

[3DVagabond]

Except for being a terrible analogy since the "uber" switch does not alter visuals or optimizations.

The closest analogy is if Nvidia decided to offer and promote a software turbo button in their drivers that upped boost clocks and perhaps fan speed and voltage beyond the shipped mode and ensured this was covered by warranty. Let's say getting this mode running was as simple as moving your mouse over to a desktop or taskbar Nvidia icon and clicking perhaps twice. Would you expect reviewers to not test that mode?

Actually it's not even all that. It simply allows the fan to spin up faster.

 

[3DVagabond]

What a ridiculous rant. Nobody is condemning extra options.. Uber mode is fine, I have no problem with it as an extra option for consumers to increase performance.

The only problem is, how do you resolve it when it comes to reviews, which are supposed to be FAIR and impartial.

That's why reviews are conducted in default mode in the first place, to impose fairness. There is always extra performance to be gleaned for those that want it whether through overclocking or adjusting settings, but that should not be the focus of a baseline review, the purpose of which to showcase out of the box performance.

Uber mode does not increase performance. Quiet mode, as it's name suggests, simply reduces noise by limiting fan speed.

 

[railven]

Re read my post. I didn't say a reviewer can't give a recommendation. I said an unbiased review should be very careful in how to recommend, or recommend against a product. The language used in the reviews was not carefully worded. Frankly, constant focus of the reviews on noise was not necessary.

Everyone agrees the cooler could be better which would in turn make the GPU more quiet. People came to that conclusion after reading countless other reviews that did not smash the reader over the head with a "NOISE BAD" hammer like Ryan's review at Anandtech.

He stresses noise, clearly that was his issue. How does that make him bias? If anything, it makes him bias towards not_so_loud_cards, which the R9 290 is not.

You guys need to look up what bias means.

I've seen reviews condemn a card based on price (hey, that's also a subjective measurement, is it not?), or heat, or noise. Basically the whole product is subjective, the only objective info is the performance.

Where Ryan gaffed isn't even in the R9 290 article, and I don't know if some of the posters in this thread know or not, but he did test Quiet/Uber modes for the 290X. His mistake was throwing out Uber results in the 780 Ti review.

There is nothing wrong with the R9 290 article.

 

[Vesku]

Actually it's not even all that. It simply allows the fan to spin up faster.

I translated it to Nvidia's boost methodology, they may in the future adopt a target clock ala 290(X) powertune but I'm not a soothsayer.

It's already been pointed out that Anandtech turns DP on for Titan compute so the lack of 290X "uber" comparison in the actual 780Ti review text, not talking graphs, is an odd stance to take.

Even the graph data for "uber" was added in after some community feedback, yes?

 

[3DVagabond]

Good catch.
What is wrong with this chart and why?

How will nv entry compute card look with your policy in place and why it doesn't look this way?

I agree with the way this chart was done. It would be completely unrealistic to not engage DP mode on Titan while testing DP performance. The results given would be worthless.

I also agree though that it's not the default setting so it appears to be a double standard.

 

[Wall Street]

Did I seriously just read that? I can't read this with a straight face, that the AMD noise issues are overblown. At 40%, yes, it is quiet. At 55%? If you think 55% or anything past 45% is quiet on an AMD reference shroud, you're basically being disingenuous. I've owned 3 generations of ATI cards and I know what they sound like, and they're all identical under the hood - even the 290X shroud is the SAME as the 5870, 6970, and 7970 shroud. IT IS NOT quiet past 45%, it is especially not quiet compared to any Kepler shroud, and it is VERY VERY stupid that to get a quiet card that you must lose 20% performance.

Saying that Ryan was overblown on the issue is not saying that the fan is quiet. Stop making straw men and putting words into people's mouths.

The fan noise needs to be evaluated in the context that this card is $100 less ($250 less compared to one week ago) than the GTX 780 and is closest in price to the GTX 770 from nVidia which it is much faster than. It performs close to nVidia's $1000 card and you wouldn't find Ryan saying Titan is a ripoff (it is, but that shouldn't be said in a review because it is too subjective).

When I first really got into custom computing, Delta black label fans were popular and the trade off of being able to get a much cheaper processor to be faster than the most expensive stock offering in exchange for heat and noise was worth it for many enthusiasts on a budget. The R9 290 is similar in that it is clearly the best perf/$, even if you have to spring for an aftermarket cooler or wait for a $20 more expensive DCII/Windforce/Twin Frozr or accept the noise.

Frankly, if AMD didn't release the 290, there would be no card available even close to that price/perf. nVidia had to drop the GTX 770 price by $70 from the same $400 price point and had to cut $150 from the GTX 780 price so nVidia certainly doesn't consider the 290 as poorly as Ryan does. I am astounded that Ryan can so easily ignore price differences of as much as $600 (2.5x total price for Titan) in cards that perform similarly (although he obviously doesn't pay for review cards out of pocket).

 

[Wall Street]

I'm talking manual fan speeds. You know, there have been utilities for adjusting manual fan speed to comfortable levels, for YEARS.............and there hasn't been a GPU that tanked your performance for using a reasonably quiet fan speed. Not the 5870, not the 6970, not the GTX 280, etc......

If you manually adjust just about any card low enough, it will throttle or fail.

 

[3DVagabond]

Well as you explained yourself you dont use the uber because you might miss the delivery man?

No. Because he might break a nail. Comprehension fail /joking

 

[3DVagabond]

Honestly, its because its TWICE the workload to bench the same card at two different conditions.

It's NOT a reviewer friendly option with these bios switches AMD is doing. In future, they need to drop it and go full reliance on powertune boost, that with a $10 more expensive reference cooler will solve all their issues.

So indeed, AMD needs to be sent a message, because from their prior history, they simply don't get it unless they are told. Its the same story over again, AMD, awesome engineers, horrible management. Who thought it was wise to put essentially the same 7970 reference heatsink on it? That person needs to be fired.

Probably the same people who decided they wanted the 290X to undercut the 780 by $100 and that was required to do that.

I'm not sure where you get the $10 value of implementing a new cooler. Titan's cooler did not cost $10. Keep in mind that there are other costs besides the BoP on a new cooler design. Look at what nVidia did with Titan. They limited it to only reference cards and charged $1000 for it. I'm sure that the cost of the cooler, total cost not just BoP, had a bearing on the reference design only. It likely also influenced the $1000 price tag a bit as well.

 

[TerryMathews]

I agree with the way this chart was done. It would be completely unrealistic to not engage DP mode on Titan while testing DP performance. The results given would be worthless.

I also agree though that it's not the default setting so it appears to be a double standard.

The correct answer, I think, is to create some consistent test criteria and apply them across like cards regardless of "default settings" or the position of any switches/jumpers when you take the item out of the box.

 

[3DVagabond]

GOOD. This needs to be done. I've said it before but noise is only part of the problem, the other MUCH BIGGER problem is throttling that results from that crap cooler.

AMD needs to understand that this is not acceptable - too much noise is not acceptable, and throttling with quiet acoustics is REALLY NOT acceptable. I personally hope that other review websites send this message as well. The fact that the 290 and 290X cards require a 20% drop in performance for a quiet fan profile is downright RETARDED. Come on, are you kidding me? How can anyone defend this crap? AMD should have made a better shroud. Period. I hope more websites make this message loud and clear to AMD that they screwed up, people do care about reference blower quality because not everyone is able to use an aftermarket design.

Good LORD. You people need to stop acting like AMD is the bullied victim here. THIS IS THEIR FAULT. PERIOD. I praised the 7970 back in the day because I thought it was a great GPU but it also DID NOT require an insanely high fan profile for good overclocked performance. I didn't lose 20% performance at 40% fan - I could overclock to 1100 *WITH* 40% fan. The GPU landscape has changed. This is not and should not be acceptable, and only if consumers and review websites hammer that point repeatedly will AMD get the message.

You are still completely ignoring the fact that the 780 also throttles with the very expensive Titan cooler. These cards will not operate without reducing clocks with a 2 slot blower style cooler, even a really expensive one like they use on the Titan. To operate at their full potential they require an open air design or possible an over sized blower cooler. This is what the AIB's offer us and AMD is happy to allow them to do that.

 

[blackened23]

You are still completely ignoring the fact that the 780 also throttles with the very expensive Titan cooler. These cards will not operate without reducing clocks with a 2 slot blower style cooler, even a really expensive one like they use on the Titan. To operate at their full potential they require an open air design or possible an over sized blower cooler. This is what the AIB's offer us and AMD is happy to allow them to do that.

"They both throttle". Yeah okay if you say so. One throttles by nearly 300mhz and another barely throttles even with vsync off at 100% GPU load -I've used kepler cards for quite a while now and the throttling is far less of an issue and does not cause a 20% drop in performance. In my personal use, none of my kepler cards have ever throttled more than 2 bins (26mhz) in real world gaming, and that is after 4-5 hours of warmed up use.

Meanwhile, you have the 290X with 40% fan speed throttling by 300mhz. Kepler's boost has variances but it is so minutely small it isn't worth noting, and it doesn't cause more than a 1-2% performance drain at MOST. Furthermore, the boost speed avertised isn't a "holy grail" type of speed, all kepler cards boost WELL PAST the advertised speed listed on the box. For example, the GTX 780 advertises a 914mhz boost, nearly every person i've spoken to boosts past 980 easily. Further, my SC ACX 780 advertises 1020mhz, and it boosts around 1100 out of box without overclocking, and all Kepler GPUs are similar.

The differences with AMD's powertune are:

1) throttling by 200-300mhz as opposed to Kepler's meager throttling of 1-2 bins in the worst scenarios.
2) the boost is an "up to" speed while Kepler's boost is guaranteed and all Kepler GPU cards boost WELL PAST their advertised speed out of box, without overclocking.

You can say both cards have the potential to throttle, but I guarantee one of these cards don't throttle to lose 20% performance. And that ISNT the 290. This is the single reason that AMD's decision to use that cheap shroud is so disappointing - I don't know of any GPU that had this type of variance for low fan RPMs. I know the 5870, 6970, and 7970 sure didn't and I quite liked all of those cards - all of those GPUs could be used at reasonably quiet levels WITHOUT sacrificing performance.