Graphics The new mag has thrown a spanner in the works

Hi guys,

Feel like I should jump in here before this turns into a hate-spouting circlejerk based on misinformation and/or misunderstanding, as while I did not write the reviews personally, I did do all the testing and had a large say in the content of the reviews and calculated the scores for the group test in question.

Firstly, I want to make my position clear on the 970 specs thing, and I pretty much speak for the mag here too since I am almost fully responsible for all of its GPU coverage. I actually believe it is an almost non-issue. I was one of many reviewers who lavished the 970 in praise on launch day, and I stand by that praise. It is a fantastic card, it launched at a great price point (surprisingly so for Nvidia) and had efficiency unseen. I *do* think it is unacceptable that information was miscommunicated to press, but as has been pointed out it doesn't really change anything. If you bought the card prior to knowing it, I don't think you were undersold; nobody really buys cards because of ROP count, L2 cache or memory segmentation. They buy for performance, which doesn't change and hasn't been show to be diminished in any way that is non-negligible since it all came to light. I considered blogging about it on bit-tech when it all erupted, but honestly that's all I have to say about it. In summary: meh.

These feelings are reflected in the review, where this is a single small paragraph that explains the issue for those who might be confused. This is summarised by saying "These are both disappointing issues, but we’re concentrating on performance." Before this paragraph there is only a discussion on the other specs, and afterwards only on the test results. It is never mentioned again - it is not mentioned in the conclusion, the verdict and it is in no way taken into consideration in the scores, which reflect 5 things only: performance (in 3 different resolutions), efficiency and bang per buck. There is no awarding or removing points based on telling people the correct specs.

So, supermuchurios, to say that the score dropped because "of Nvidia lying" is a TOTAL misrepresentation of the entire review, and it is actually damaging for you to do this so carelessly: Cei has now taken it upon himself to call it "ridiculous" and "ludicrous", to accuse the mag "jumping on some internet rage bandwagon" and questioning the author's integrity - he has freely admitted having no access to the mag yet has now formed an entirely negative opinion based entirely on what you said, while CrapBag has also falsely aligned the specs mishap with the new, lower score. I have no issue with criticism of my writing or the reviews and features I'm a part of, but please do not misrepresent what we are doing like this, especially in so public a forum.

So the question remains: why did the score drop? The answer is that for graphics card group tests we focus entirely on price vs performance, with the specific aim of fiding out what cards are suitable for smooth gameplay at certain resolutions (in this case 1080p, 1440p and 4K) as we feel this covers what the majority of people are looking for, and it is grounded in reference to real-world usage. It is a completely different scoring process from the standalone review, and it's designed to provide a snapshot of the market (as up to date as possible in a print mag) based on price vs performance, and it is explained in detail in the How We Test section. Yes, the 970 dropped to 76%, but the highest scoring card was the R9 290 with 80%, so to say we are pissing on the 970 is again absurd. It is made clear in the review that it is a great card, but the 290 is cheaper and offers better performance. Yes, with less efficiency, but that is also a component of the scores. And before anyone mentions the R9 290/290X's reference cooler being crap, I totally agree. That is why the tested cards for those models were third-party cooled, and the pricing based on those cards, not reference cards as with the other models on test. I would never recommend buying a reference 290 or 290X, so they were not considered in the testing or score calculations.

Sorry for the long reply but I have been put on the defensive here. As I said, please do feel free to criticise my work, but at least have the decency to base this on fact, and be more wary of those who attempt to do otherwise.

Matt

 

Kovoet, I am sorry the group test couldn't help you fully decide on a card to buy, but you are right that most are good now, and performance is generally in line with what you have available to spend. Still I hope you're able to see that there are certainly a few better models than others. Once you've decided as well, be sure you're not paying ridiculous amounts for one specific brand e.g. Asus, MSI, EVGA etc. as then you might be able to move up to a better GPU.

CrapBag, it may seem counter intuitive, but the two 970 reviews are not directly comparable. The scoring procedure and the reasons behind it are explained in the mag. The cards are in fact all calculated against each other, not just awarded at the hest of the reviewer. It may have dropped in percentage terms a lot, but the scoring system is so so different that this really doesn't mean anything. As I said, the highest score now is 80%, making 76% a VERY good score in the context of the group test, not a fall from grace. And the reasons for the 290 beating it are explained in the reviews and by the scores. I'm not sure what's still unclear but I can't be of much more help unless you're more specific about what you don't get.

As for the AMD comments, I did not comment because I am not really the person to do so, unlike with GPU coverage. Nevertheless.. for CPUs, I do not typically cover them at all outside of the Labs feature. There has not been an enthusiast CPU platform launch from AMD since I took over labs unlike with Intel (which has had Z97, X99 etc.). When there is, I'm sure we will do a full roundup of the new CPUs and motherboards as we did with those. For standalone reviews, I do not typically review CPUs (though fwiw the two I have done were both AMD, as I covered the Kaveri and Richland launches), so AMD won't approach me with those products. As for coolers, the last air cooling roundup (Issue 132) included FM2+ results, which took into account and compared the coolers both on cooling results and AMD fittings. The liquid cooling roundup in Issue 138, because as was explained we don't think there is a large market for people strapping liquid coolers to FM2+ CPUs. Yes there are other AMD platforms, but we ddidn't do cooler testing on those since atm at least we don't recommend them in any of our builds etc.

I have also made Antony (who covers more CPUs and coolers) and Ben (editor) aware of this thread and the issues raised in case they want to add anything further for you

 

Thanks for the apology I'd be happy to hear your thoughts if/when you have read the content.

Appreciate the feedback; I understand what you mean there, and I agree it could be confusing. However, it's just the nature of group tests; they are self contained snapshots of a certain market at the time of writing. In this instance, we really focus on price vs performance, as card features, power draw and more all vary greaty between specific models, and there is clearly no feasible way to test every single model. We want users to be able to decide what card they should buy with a set amount of cash, and what resolution they can expect to play comfortably at with high settings. We do endeavour to clarifiy that the scoring system is different and shouldn't be compared to standalone reviews, but clearly there has still been some confusion. I will make more of an effort in this regard in future tests and flag it up with Ben too to see if there's anything more we can do.