Last night we published our Radeon HD 6870 and 6850 review. In it we made a decision to include a factory overclocked GeForce GTX 460 from EVGA (the EVGA GeForce GTX 460 FTW). For those who aren't aware, NVIDIA has allowed a number of its partners to ship GTX 460s at higher than stock clock speeds. A practice that has been done in the past. The cards are available in retail with full warranties.

A number of you responded in the comments to the article very upset that we included the EVGA card. Even going as far to accuse us of caving to NVIDIA's pressure and demands. Ryan and I both felt it was necessary to address this front and center rather than keep the discussion in the comments.

Let's start with the obvious. NVIDIA is more aggressive than AMD with trying to get review sites to use certain games and even make certain GPU comparisons. When NVIDIA pushes, we push back. You don't ever see that here on AnandTech simply because I don't believe this is the place for it. Both sides (correction, all companies) have done nasty things in the past but you come here to read about products, not behind the scenes politics so we've mostly left it out of our reviews.

NVIDIA called asking for us to include overclocked GTX 460s in the 6800 series article. I responded by saying that our first priority is to get the standard clocked cards tested and that if NVIDIA wanted to change the specs of the GTX 460 and guarantee no lower clocked versions would be sold, we would gladly only test the factory overclocked parts. NVIDIA of course didn't change the 460's clocks and we ended the conversation at that. We gave NVIDIA no impression that we would include the card despite their insistence. The decision to include the EVGA GeForce GTX 460 FTW was made on our own entirely.

We don't like including factory overclocked parts in our reviews for reasons we've already mentioned in the article itself. This wasn't a one off made for the purpose of reviewing only, it's available from online vendors and a valid option from a price comparison. Furthermore it presented us with an interesting circumstance where the overclock was large enough to make a significant impact - the 26% overclock pushed the card to a performance level that by all rights could have (and should have) been a new product entirely.

From my standpoint, having more information never hurts. This simply provides another data point for you to use. We put hefty disclaimers in the article when talking about the EVGA card, but I don't see not including a publicly available product in a review as a bad thing. It's not something we typically do, but in this case the race was close enough that we wanted to cover all of our bases. At the end of the day I believe our conclusion did just that:

At $179 buy the 6850. At $239 buy the 6870 for best performance/power. If you want the best overall performance, buy the GTX 470. However, as long as they are available the EVGA GeForce GTX 460 FTW is a good alternative. You get the same warranty you would on a standard GTX 460, but you do sacrifice power consumption for the performance advantage over the 6870.

We were honestly afraid that if we didn't include at least a representative of the factory overclocked GTX 460s that we would get accused of being too favorable to AMD. As always, this is your site - you ultimately end up deciding how we do things around here. So I'm asking all of you to chime in with your thoughts - how would you like to handle these types of situations in the future? Do we never make exceptions even in the case of a great number of factory overclocked cards being available on the market? Do we keep the overclocked comparison to a single page in the review? Or does it not matter?

And if you're worried about this being tied to financial gain: I'll point out that we are one of the only sites to have a clear separation of advertising and editorial (AnandTech, Inc. doesn't employ a single ad sales person, and our 3rd party sales team has no stake in AT and vice versa). The one guarantee that I offer all of our writers here at AnandTech is you never have to worry about where your paycheck is coming from, just make sure you do the best job possible and that your conclusions are defensible.

If we've disappointed you in our decision to include the EVGA FTW in last night's review, I sincerely apologize. At the end of the day we have to maintain your trust and keep you all happy, no one else. We believed it was the right thing to do but if the overwhelming majority of you feel otherwise, please let us know. You have the ability to shape how we do things in the future so please let us know.

Whether you thought it was an issue or not, we'd love to hear from you. I do appreciate you reading the site and I want to make it better for you in the future.


Take care,

Comments Locked


View All Comments

  • vedye - Wednesday, October 27, 2010 - link

    To summaries your comments:
    1. anyone disagrees with you are fanboyz/trolls. Despite the availability of the included FTW card is extremely limited, there are people like you that loves seeing it comparing to the 6800. Love the bars.

    2. You love anandtech this way and if they do less like this, you'd be disappointed.

    Well, one thing for sure is that Anandtech had been famous NOT for being a biased, or just another review sites. I for one was attracted to this site for its in-depth knowledge and fairness position. For the past few reviews of Anandtech caving in to Nvidia's pressure? Yes it will attract people like you, but it surely has lost my respect, and probably many more like me.
  • mapesdhs - Wednesday, October 27, 2010 - link

    Yesterday, in response to one of my posts, someone insisted that site reviews were not finding
    the FTW to be as good or better than the 6870 (which is nonsense of course). I mentioned that
    I'd found another such review, so feel free to have a look - it's on techradar (just add the dot com),
    search for 6870, check page 2. Out of seven gaming tests, the 6870 is faster for only one of them,
    and for half the other six tests the FTW's margin over the 6870 is substantial.

  • Touche - Wednesday, October 27, 2010 - link

    Techradar? LOL! And their game selection is awesome. What a joke.

    You can check some of the more reputable sites, maybe Overclockersclub, and see how FTW looses in most 1920x1200 bencmarks and even more in 2560x1600.
  • kc77 - Wednesday, October 27, 2010 - link

    I commented about this on another site and I'll try to be a precise as possible with the problem of allowing Manufacturer A to slipstream cards into Manufacturer B's hardware launch.

    In and of itself providing more data points is not a problem within a review. People complaining about that are really missing the REAL problem of doing this.

    The first problem is that when you allow a manufacturer to change your test matrix by inflating the number of data points within a hardware launch it deliberately takes away resources from an article that would otherwise included more information. While you were able to include CF results, many other sites which did the same thing as Anand did (and included more Fact OC'd cards) were not. Many were not able to include info on new AA modes, even when their reviews were late. Many sites barely went into the display configuration changes on these new models, or if they included information it was brief ...really brief. I don't think ANYONE covered Eyefinity changes or improvements or benchmark data. Considering this was a key feature of Radeons it should have been covered ..... in detail. Many sites gave almost equal billing to cards which were on the market for at least 3 months or more to the point that reading this "launch review" almost seems like a GPU shoot out review rather than providing in-depth information on new hardware, which is what launch articles are for.

    The second problem is as I mentioned before. It is odd, rare and not normal for a GPU shoot out review to appear within a launch review. By doing this it left NO ROOM for Fact OC'd cards to appear from ATI within the review because that wasn't the point of launching new models. Fact OC'd cards are likely to appear, actually they have of 6870 cards...yet considering a shoot out review was already done with 460 models, will another Fact OC review appear from ATI so we can see those data points? You may, but lesser sites might not as they might not have the same access to the 460 cards as they did on date of the launch of 6870/6850 which DOES NOT give us more information to make informed choices, it gives a less or a distorted view of what is out there.

    The third and final point is that this just stinks from a ethical or journalistic point of view. Not because I like ATI over Nvidia, but because it creates a precedent that NO SITE IS LIKELY TO REPLICATE GOING FORWARD. Meaning will you or other sites include Fact OC'd cards when the 580 launches? How about the next review? Or the one after that? The moment you don't it breaks the farce that "we did this to bring you more information" because you won't be able to explain the reason to include more information in one review, while not doing it the next.

    As a reader of tech sites I should be able to come to Anandtech, TR, Hard, etc and not question the validity of the results. I should be able to look at the numbers and say " well they are what they are" knowing and believing that every effort was taken to provide a review that is of the sites making. Not Nvidia's or ATI's. When you allow manufacturers to change your test matrix how are we to know that the 480 wasn't a cherry picked model, or the 5870 wasn't cherry picked? By doing this it opens up a pandora's box that would have remained closed if we knew that your standards were upheld and were free from being tainted. Now we have to question results when we wouldn't have before. That's the problem with doing this.
  • Nate007 - Thursday, October 28, 2010 - link

    This whole controversy could have been avoided by AnandTech by doing a stock comparison of all the cards and then a ANOTHER time later on they should have done another article on the cards OC.
    Hardware Canucks decided to do just that and only reviewed and benched STOCK cards as is there policy NO Matter what AMD or NVIDIA wants. They also stated that they will follow up with a review of the cards OCed on the next follow up.
    I don't want a review of the product OCed if I did then I would ask for it and or make it a point to let your Manufacture know about it.
    It;s the same way that I wouldn't want a review on a product (ex :.. Iphone ) Oced and then compared to another smart phone of equivalent value /features and what not. Just wouldn't tell me the truth of the product as sold my the Manufacture.

    The problem here is when you cross the line and start doing reviews of products that are not in STOCK reference state then you run into many other issues with regards to credibility , simply because that is the World we live in.
    Stick to stock designs and avoid all the neg feed back next time. Besides no one said you couldnt do a separate OC review later.
  • kc77 - Thursday, October 28, 2010 - link

    Exactly and that's my point. NO MANUFACTURER SHOULD BE ABLE TO CHANGE THE TEST MATRIX. It left no room for rebuttal and alot of it could have been avoided by simply posting a second review after the launch date which would have ensured that a full review was done on new hardware. The 460's were 3 months old and could have been reviewed before.
  • GinandTonic - Thursday, October 28, 2010 - link

    OK, I can't resist. Both of us must have way too much time on our hands, lol.

    I'll summarize what I hear (metaphorically) you saying about the "problem" of one manufacturer slipstreaming into another's launch.

    1. Including that data takes away from the resources/space available for the review of the launch manufacturer's product.
    2. It's not "normal" and isn't fair to the launch manufacturer who hasn't yet provided/developed oc cards, therefore we actually have less/distorted information.
    3. It sets a precedent that will be hard to replicate going forward, (conceptual leap coming up) and if they can't that will prove that providing us data on the 460oc this time was a "farce."
    4. This all makes you question the actual validity of the hard numbers/data reported in the review.

    I may be overanalyzing this, but I think that I'm also hearing a subtext of not fair to the launch manufacturer.

    I agree with point 1. It is obviously true, though I would suggest not very relevent. It is not enough for me to know every detail of a specific product without some context. I want to know the important features, and the actual performance/value compared to other products that are also competing for my not unlimited dollars. I think as a consumer that any review (launch or otherwise) should always be a little bit of a shoot out or else it becomes meaningless.

    I can't agree with any part of point 2. I place no value on "normal," or on how anyone else does something. I also think that it's unfair to me to withold data from me on an available product until other manufacturers have a chance to catch up. I don't see how I have less information by having more information on what is available at the time of the review. That less is more thing may apply to Bauhaus architecture, but in most areas of life, less is less and more is more. A review site should have no interest in being fair to a manufacturer at the expense of the people they actually work for -- their readers.

    Point 3... I hope that Anandtech and others will always provide me some context/comparisons of similar (performance/price) available products to help me judge the relative value of the launch product. I don't think it will really be all that hard to do.

    Point 4, ok. If that's how you feel, though you lost me on the connection.

    There seems to be a pervasive feeling in some of the comments that Nvidia did something bad by trying to get reviewers to promote their product and by trying to create a competitive product to ATI's latest launch cards.

    Apropos dealing with the press, someone once told me that there were no inappropriate questions, only inappropriate replies. Both Nvidia and ATI try to influence reviewers, and good reviewers resist those efforts. Resisting means writing straight-up reviews from the standpoint of their readers. It does not mean refusing to review a product their readers are interested in because OMG, the manufacturer also wanted it reviewed.

    Second, I am not going to be angry at either manufacturer for trying to make a better product for me at a better price. I actually think that's a good thing and I would like to encourage both of them to continue working hard for my dollars.

    Finally (for you brave/bored souls who stuck with me) I want to reply to the comment suggesting less gin and more tonic. Less is less and more gin is more of a good thing. Cheers
  • lakrids - Thursday, October 28, 2010 - link

    What happened to you? Was my last question too hard for you?

    Why was nvidia allowed to have factory overclocked cards in the 6800 review, when ATI wasn't allowed to have factory overclocked cards in the gts 450 review?

    6800 review: Nvidia was allowed factory overclock, but ATI wasn't.
    450 review: Nvidia was allowed factory overclock, but ATI wasn't.
    460 review: Nvidia was allowed factory overclock, but ATI wasn't.

    Do I need to reiterate that we had access to lots of factory overclocked 5770 cards, as well as factory overclocked 5750 cards up to the 450 review?
  • kc77 - Thursday, October 28, 2010 - link

    It's OK if you can't resist neither could I. :)

    1. It looks like you got this point. However, there were many things that were key features of the card that were not addressed that people WILL use. Eyefinity benchmarking / review to name one, considering these cards now support an additional monitor. How does it perform? How does it perform compared to the 460? You can't answer it and neither can I but it would have been nice to know don't ya think?

    2. Number 2 actually does withhold data because there ARE FACT OC 6870's. Some substantially overclocked. If you didn't know this well..... I guess the review didn't include enough data points. The point of "fairness" that your extrapolating has really nothing to do with treating either manufacturer with kid gloves. It's about making sure that the review is giving an accurate diverse representation of what's available. So when you see Fact OC cards from one manufacturer and not the other that's not necessarily an accurate snapshot. In fact that's the opposite, which is why you don't have GPU Shootouts slipstreamed into a GPU launch articles. Those would normally be two separate review articles.

    3. Point number 3 is no kind of conceptual leap at all. You are either providing additional data points to hopefully provide a better picture to your readers or you're not. Including the 460 FTW in and of itself isn't a bad thing. However, doing it one time only basically says it's only important one time only. If that's the case then the obvious question is, "why is that?"

    4. The feeling isn't pervasive it's pretty front and center. A hardware journalist allowed a manufacturer to directly change/alter their test matrix. To be honest AT didn't cave to the point that others did. Some provided so many OC'd 460's that you could barely tell it was a launch article. That being said, stealing one cookie from the cookie jar while others took eight doesn't mean you get off.

    Now as far as this.....
    "Both Nvidia and ATI try to influence reviewers, and good reviewers resist those efforts. Resisting means writing straight-up reviews from the standpoint of their readers. It does not mean refusing to review a product their readers are interested in because OMG, the manufacturer also wanted it reviewed."

    This really is a straw man argument. No one is saying that AT should NOT review Fact OC'd cards. There's a place for these cards and for years they always appeared in a "shoot out" type of article or a straight up review on the Fact OC'd card itself.
  • Nate0007 - Thursday, October 28, 2010 - link

    I understand and you do make valid points with your post.
    I totally agree with you especially your 3rd point.

    Like I said it creates way to much " what if's " and in the end it is not a Apples to Apples comparison.
    I know why ATI or NVidia would not have an issue with it so long as it is there product that is put in a better frame of light so to speak, but for a Review sight such as AnandTechs I just don't see the point.

    There are already too many review sites are starting to do this and this is what is causing all the uncertainty. I used to frequent Toms H allot in the past but personally I feel they are also now creating reviews that are confusing and not providing the same reviews that they were in the past.
    This causes people to start questioning what is going on and why it is done that way.
    Reviewers MUST be neutral and just supply the facts as is. Just provide the data A to A , B to B
    Then let the public decide.

    Lets hope it doesn't happen here too.

Log in

Don't have an account? Sign up now