tweaktown review GTX660Ti

[BenSkywalker]

You think there is some AMD conspiracy theory here?

They did what AMD asked them to. Rational people would normally consider that fair.

That doesn't explain why the 7950 wasn't overclocked at all vs. overclocked 660Tis in the last section.

http://www.anandtech.com/show/5818/nvidia-geforce-gtx-670-review-feat-evga/19
http://www.anandtech.com/show/5625/...-7850-review-rounding-out-southern-islands/19
http://www.anandtech.com/show/5458/the-radeon-hd-7970-reprise-pcie-bandwidth-overclocking-and-msaa/2

Your assertion is that they need to change their review policy to champion your board of choice. And you see this as reasonable. I guess that really sums up this thread in a nutshell.

 

[RussianSensation]

My board of choice?

That's hilarious. So you are accusing me of bias already? Let's stick to the topic and not point fingers or assume any personal connection to the product. I didn't have a problem with EVGA GTX460 FTW in the 6870 review either since it was a product the consumer could buy and it was in some ways better than the 6870. If you are trying to paint me as some AMD fanboy, you are strong mistaken.

So what you are saying is I can't even question the "authority"? If I provide a suggestion it's automatically disregarded because it's viewed as AMD biased or because you the "Authority" decided it's unreasonable? I didn't know the world revolves around Rules and some Status Quo. This isn't the military in case you haven't noticed. We are entitled to voice our opinions and in my opinion it's not reasonable to test a reference $350 7950 against after-market 660Tis when cheaper 7950s are for sale that have cooler and quieter after-market coolers.

It's not unreasonable to include after-market 7950s in the same review as after-market 660Tis especially given the $20-30 price delta on a $300 card. Other reviewers such as TechReport have done that.

You seem to want to jump at me for trying to change some policy as if you are employed by AT, when all I am saying is that more fair and reasonable reviews would be better. If you think having 1 brand OCed in a review and another brand in another review is convenient, I can assure you you are in the minority on our boards. We can even have a poll and I guarantee you'll lose it. People would find it far more convenient to have stock vs. stock and OCed vs. OCed cards in the same review. It's fast and easy to see. The review would cover both enthusiasts and users who won't overclock without having to jump hoops to find such data.

Regardless, at the very least after-market 7950s should have been included since they are available for sale in the world today for $350 or less, which puts them in some competition with the 660Ti.

 

[tviceman]

You think there is some AMD conspiracy theory here? I think even in AMD's reviewers they should overclock AMD and NV cards. In this case not only did they ignore overclocking potential of the 7950 but they used after-market 660Ti vs. reference 7950. Even TechReport included the MSI TF3 7950 to at least have 1 after-market 7950. That's being objective 101.

Anandtech rarely (very, very rarely) uses non-reference versions of cards in reviews of other cards (i.e. if they are reviewing SLC4180-XT they will only include reference versions of pl0-G3R-SX for comparison). The only time I can think of in the last several years of card reviews where they included a non-reference version of a card in another card's review was with the hd6870 (when they included the EVGA gtx460 FTW). Such a fuss was made about it being biased and unfair that I'm pretty sure they decided to not include competitor's custom cards in a new reference card review. AT's track record in this behavior has been very consistent before and after that one incident. I don't see why all of sudden it's considered shoddy reviewing.

(BTW I'm not arguing the value of the gtx660ti or the hd7950, I'm just pointing out how AT's reviewing behavior has changed very little in the last few years and has generally been good enough. If anything, this just shows how many holes and messed up AMD's handling of GCN has been).

 

[Ryan Smith]

Hi guys. My ears are on fire so I thought I'd drop in.

The subject of factory overclocked cards comes up periodically, so this isn't new for us.

First and foremost, as a matter of editorial policy we will never ever cross-compare factory overclocked cards. Factory overclocked cards are short-lived SKUs and it's not unusual for them to be suddenly discontinued, or worse have their design changed mid-way through their run. Furthermore their pricing is very erratic; this is less the partners' doing and more the vendor's (e.g. Newegg), but it means the value proposition can change drastically overnight. Whereas reference type cards, especially given the fact that you can lump them together given their interchangeability, are very consistent when it comes to day-to-day pricing.

Anyhow, this doesn't mean we won't look at factory overclocked cards, but it does mean we do so with restrictions. We look at factory overclocked cards to compare them to other cards within a specific model family, such as looking at factory overclocked 660 Tis when looking at the 660 Ti. The purpose of such testing is to break down the large number of cards available within a model family in order to look at the individual cards and to evaluate them for their performance, their cooling, etc.

Fundamentally, our philosophy is very much hierarchical. We seek to first establish baseline performance information for model families (e.g. how does the reference 660 Ti perform compared to the 7950), and then we look at individual cards within those model families to figure out what the best 7950 is or what the best 660 Ti is.

This philosophy largely goes for end-user overclocking too. We'll overclock cards to give our readers an idea of what we accomplished, but the purpose of that testing is largely exploratory. It isn't to decide whether a card within a model family is better than a card within another model family because it overclocked farther (e.g. a cross-comparison).

This is because overclocking is not guaranteed, only the out of the box experience is. It is critical that we do not lean too heavily on our overclocked results, because if we make recommendations based on that, we're one very good (or very bad) overclocker away from making a bad recommendation. Overclocking is a nice bonus, but when it comes to the wide variability of GPUs you should never rely on it.

For these reasons, our principle recommendations are based on reference type cards at reference clocks. This is the baseline of performance; this is what NVIDIA and AMD can guarantee.

 

[3DVagabond]

What BenSkywalker is saying is that it's AMD's fault for setting the parameters themselves.

If AMD wants to show people how good these MSI cards are they need to get them into reviewers hands. Send them the 7950 that uses the 7970 PCB. Ask them to take the time to do Stock, max O/C at stock voltage, max "efficient" O/C (As high as they'll go before the O/C vs. voltage used drops off) and max O/C voltage be damned.

If you look back at the original 7850/7870 release reviews and you'll notice that none of the 7850's were O/C'd with voltage, but all of the 7870's were. I'll bet that was done that way because AMD didn't want the 7850 to outshine the 7870. if the 7850 was O/C'd with voltage from day one's reviews then everyone would have known how great of a card it was, right from the start. That didn't happen until they started doing reviews with added voltage on the 7850 and that's when we got the "7850 is an O/C'ing beast" thread. That's also when we started to see shortages of the 7850 DCII after Ocaholics review. Where their card basically matched a custom O/C'd 580 for ~1/2 the price.

I don't think he's honestly saying AMD is conspiring against themselves. Just that they couldn't have done a better job of screwing themselves over if they had tried to. He's pointing out how stupid their marketing is.

 

[BenSkywalker]

So you are accusing me of bias already?

Reviewers did what AMD asked them to, and you say it isn't good enough. You are saying that you are more biased towards AMD's parts then they are themselves on this particular topic. AT followed AMD's suggestions reviewing a nVidia part, and that still wasn't good enough for you. You are claiming bias yourself, I'm just pointing out that a site followed their own review policy and tested AMD's parts as they requested.

I didn't know the world revolves around Rules and some Status Quo.

Each site has their own review policies. People in this thread blasted Brent for being biased too because he didn't do in depth overclocking, even though he normally does that in a separate article. What we are seeing is a bunch of people calling reviews biased because they didn't change their editorial policy to make the 7950 look better then the product AMD sells.

You seem to want to jump at me for trying to change some policy as if you are employed by AT, when all I am saying is that more fair and reasonable reviews would be better.

You don't want fair and reasonable, you don't want anything approaching fair and reasonable. The 7870 is the part that is closest in price, and also the part that is in the same bracket as the stock 660Ti, but what you want is for them to compare it against a board in the next pricing segment up, aftermarket and overclocked. The part nVidia released versus the part AMD released makes AMD look less then ideal. That is not the fault of the reviewers, that isn't the fault of how they handle their reviews. That is AMD's fault for releasing such a part.

You expect web sites to go out of their way to cover up for AMD because they released the 7950 in a poor state(I pointed this out at launch btw, long before we knew anything about competitive parts).

People would find it far more convenient to have stock vs. stock and OCed vs. OCed cards in the same review.

Which model of which card and do you expect Ryan to update them every other week? If you know how an architecture scales for core and memory clocks of a particular configuration then why do you need to see them all listed side by side? You are rolling the dice when relying on overclocking of anything, certain parts normally have more headroom, but you don't know if you are going to get one of the lemons that need 1.25 and can't hit 1GHZ or if you are going to get one of the ones that are pushing close to 1.3GHZ with a minor bump in voltage.

Is it the reviewers job to spend months going over all of these to appease your desire to market this card? Does Ryan need to spend all of his money to market the 7950 to your approval? Did all of the OEMs send him 7950s to compare to the 660Ti(clearly you are under the impression that he owes his time to the interest of marketing the 7950).

That is exactly what it comes down to. People in this thread know how the 7950 scales, they know how the 660Ti scales. What you want to see is the reviewers marketing the part you have so much vested interest in, in a better manner.

Normally when a new part comes out we get benches, see how it overclocks and where it slots in to the normal lineup. We got exactly that with the 660Ti. People now throw a fit because reviewers aren't trying to cover up for how AMD ships the 7950. All of the games that are normally benched were benched. They included MSAA like everyone wanted, they even had SSAA benches in there, they had the games that heavily favor AMD's architecture, they used the special BIOS that AMD wanted them to use to increase the clock rates of the 7950. Frothing at the mouth, lunatic levels of fandom are required to justify this backlash.

 

[VulgarDisplay]

Hi guys. My ears are on fire so I thought I'd drop in.

The subject of factory overclocked cards comes up periodically, so this isn't new for us.

First and foremost, as a matter of editorial policy we will never ever cross-compare factory overclocked cards. Factory overclocked cards are short-lived SKUs and it's not unusual for them to be suddenly discontinued, or worse have their design changed mid-way through their run. Furthermore their pricing is very erratic; this is less the partners' doing and more the vendor's (e.g. Newegg), but it means the value proposition can change drastically overnight. Whereas reference type cards, especially given the fact that you can lump them together given their interchangeability, are very consistent when it comes to day-to-day pricing.

Anyhow, this doesn't mean we won't look at factory overclocked cards, but it does mean we do so with restrictions. We look at factory overclocked cards to compare them to other cards within a specific model family, such as looking at factory overclocked 660 Tis when looking at the 660 Ti. The purpose of such testing is to break down the large number of cards available within a model family in order to look at the individual cards and to evaluate them for their performance, their cooling, etc.

Fundamentally, our philosophy is very much hierarchical. We seek to first establish baseline performance information for model families (e.g. how does the reference 660 Ti perform compared to the 7950), and then we look at individual cards within those model families to figure out what the best 7950 is or what the best 660 Ti is.

This philosophy largely goes for end-user overclocking too. We'll overclock cards to give our readers an idea of what we accomplished, but the purpose of that testing is largely exploratory. It isn't to decide whether a card within a model family is better than a card within another model family because it overclocked farther (e.g. a cross-comparison).

This is because overclocking is not guaranteed, only the out of the box experience is. It is critical that we do not lean too heavily on our overclocked results, because if we make recommendations based on that, we're one very good (or very bad) overclocker away from making a bad recommendation. Overclocking is a nice bonus, but when it comes to the wide variability of GPUs you should never rely on it.

For these reasons, our principle recommendations are based on reference type cards at reference clocks. This is the baseline of performance; this is what NVIDIA and AMD can guarantee.

I think something that would paint a true picture in the overclocking landscape would be to overclock all the cards to the lowest percentage overclock achieved. So if card A get a 25% overclock, Card B gets and 21% overclock, and Card C gets a 19% overclock, then all cards would receive a 19% overclock from their stock clock speed.

It would allow your readers to easily see the true performance of the cards when overclocked, and it wouldn't hurt the sample with the least overclocking potential. Since overclocking is not guaranteed this would be an objective way to highlight the overclocking potential of cards.

 

[RussianSensation]

First and foremost, as a matter of editorial policy we will never ever cross-compare factory overclocked cards. Factory overclocked cards are short-lived SKUs and it's not unusual for them to be suddenly discontinued, or worse have their design changed mid-way through their run.

Then why would you include factory pre-overclocked 660Ti in a 660Ti review if you think factory pre-overclocked SKUs are short-lived? Using similar uncertainty that factory pre-overclocked cards may disappear, wouldn't it mean the existence of factory pre-overclocked 660Ti's is just as uncertain as the existence of factory pre-overclocked 7950 cards, or other cards, etc? Or does the 660Ti get a pass because NV allowed AIBs to not abide to reference 660Ti clocks? The 3 660Ti cards in your review tested along-side the reference 660Ti are not abiding to factory default clock speeds. By that account you are placing faith that such SKUs may be purchased by a consumer for a foreseeable short-term future. In that case, the consumer seems to face a similar uncertain with after-market 7950 cards.

How do after-market 7950s perform relative to after-market 7950s? We don't know since they were never tested using latest drivers around GTX660Ti's launch. The consumer has that missing information that has to be filled by only 1 card: HD7950 reference with GPU Boost. Is that reasonable since there are many other 7950s available for sale on Newegg, Amazon, NCIX, etc.?

Anyhow, this doesn't mean we won't look at factory overclocked cards, but it does mean we do so with restrictions. We look at factory overclocked cards to compare them to other cards within a specific model family,

In that case could you please do a follow-up article for 7950 family of after-market cards:

1) Gigabyte Windforce 3x 7950
2) MSI TwinFrozr 3 7950
3) Sapphire Dual-X 7950

I remember in the past you'd test reference cards in 1 article and then have after-market cards for that family in another article.

For example, 7800GT mini-roundup, Retail 6600GT Exploration where you looked at 11 SKUs. Many of us found these types of articles extremely useful for choosing the best SKU within a family. I wonder why you guys don't go back to this style since I am sure many people would love to find out what the best 660Ti is or what the best 7950 is, or the best 670 is, etc.

Lately it takes a while before you guys do that or you don't revisit new SKUs in a family such as the case now with 7950s where more SKUs are available and they are falling $20-30 within 660Tis.

I know that you tested 2 after-market 7950s but that was in January using old drivers and since then other after-market SKUs of 7950 that have arrived, with pricing dropping much closer to 660Ti levels. Also, in that time the 28nm node matured for 7950 cards and they are better overclockers. While overclocking is not guaranteed neither is it for 660Ti cards and it was done in the 660Ti review. Many people will never even revisit the January 31st article where you had tested some after-market 7950s but are looking for a card at the ~$300 price level.

I think it would be useful to your readers if you had revisited HD7950 after-market versions especially since the 660Ti and 7950 look to be very close competitors for the foreseeable future.

 

[BenSkywalker]

In that case could you please do a follow-up article for 7950 family of after-market cards:

http://www.anandtech.com/show/5476/amd-radeon-7950-review

Posted in January. They did test factory overclocked 7950s.

I think it would be useful to your readers if you had revisited HD7950 after-market versions especially since the 660Ti and 7950 look to be very close competitors for the foreseeable future.

Why is it Ryan's job to pay for boards and then spend his time marketing the 7950?

 

[VulgarDisplay]

I'm confused by all the negative attention being directed at Russian Sensation. He's not demanding that you buy a 7950. He's basically just noting that the review sites could do a much better job of informing their readers.

I honestly don't think they care about their readers as long as the advertising dollars are coming in.

 

[RussianSensation]

http://www.anandtech.com/show/5476/amd-radeon-7950-review

Posted in January. They did test factory overclocked 7950s.

Already addressed above. Only 2 older SKUs were tested. Sapphire has replaced that SKU with a 950mhz version, while 3 completely new SKUs from Gigabyte, PowerColor and MSI have been released since then. Also, that review uses old drivers that are completely not comparable to the 660Ti / 7950 GPU Boost review.

The consumer couldn't buy MSI TwinFrozr 7950, PowerColor PCS+ or Gigabyte Windforce 3x 7950 for $330-350 in January of 2012 since those cards weren't even available. They are now and they are in some ways direct competitors to the 660Ti/ 660Ti 3GB $339. Someone who doesn't visit AnandTech forums daily may want to know how those 7950 cards stack up against one another and how they compare to a 660Ti / 660Ti 3GB. All of those cards overclock a lot better than those early 7950 cards due to binning for them and more mature 28nm node. Plus, the early reviews with OCed 7950 uses outdated by today's standards drivers. We can't tell from AT's 7950B testing how fast a 7950 @ 1100-1200mhz is using Catalyst 12.7-12.8 There is actually enough new SKUs including the Asus Direct CUII 7950 to have a Mini-7950 Round-UP. I think it's also important now that prices are inching that much closer to $300-330 of 660Ti / 660Ti 3GB.

The landscape for 660Ti / 660Ti 3GB has changed with 4-5 new 7950 after-market cards on the scene. We haven't seen a Mini-7950 Round-up. Do you not want to see such a review of new 7950 cards? 4-5 new SKUs of 7950 would actually provide more information to AT readers, and possibly paint a much fairer picture of true 660Ti / 660Ti 3GB competitors.

I'm confused by all the negative attention being directed at Russian Sensation. He's not demanding that you buy a 7950. He's basically just noting that the review sites could do a much better job of informing their readers.

Exactly. A Tech's site is to provide us readers with more information regarding our buying choices. Now there are 4-5 new SKUs of 7950 that fall very close to the price of the 660Ti / 660Ti 3GB. It would be helpful to see for those who are considering a 7950 card, which 7950 card is a quiet one, which overclocks the best, which consumes the least power, etc. That information is NOT available since AT tested only old SKUs dating January 2012 using outdated drivers. In my opinion that is not painting a representation of what a consumer can buy today for $300-350. At the same time, at least 3 after-market 660Ti cards were reviewed thoroughly and overclocked. Since some of those 660Ti users may be cross-shopping for the best $300-350 card, I think it would add more information and not hurt the readers if a Mini-HD7950 Round-Up was revisited again. The 660Ti and 7950 are probably going to be competing with each other for the next 5-6 months.

 

[Siberian]

http://www.anandtech.com/show/5476/amd-radeon-7950-review

Posted in January. They did test factory overclocked 7950s.



Why is it Ryan's job to pay for boards and then spend his time marketing the 7950?

You nullified his rant pretty quickly. I like AT's reviews and Ryan's (long) explanation is spot on.

 

[BenSkywalker]

I'm confused by all the negative attention being directed at Russian Sensation. He's not demanding that you buy a 7950. He's basically just noting that the review sites could do a much better job of informing their readers.

Only for the 7950. There are probably a hundred SKUs floating around of various GPUs right now that AT hasn't tested, there a whole bunch of 660Tis that AT haven't tested, a bunch of 7970s that AT haven't tested, 680s/670s etc. This only became an issue when his bold predictions that the 7950 was significantly better then the 660Ti prior to launch didn't pan out in reviews. Now, it is a major factor that every review site doesn't test every SKU of the 7950. AT has done reviews a certain way for many years now, that it should be an issue now that it doesn't make a particular core look ideal due to how it ships by default seems absurd.

It's interesting that you so oppose a review of 4-5 new SKUs of 7950s since it would actually provide more information to the readers, not less and paint a much fairer picture of true 660Ti competitors.

You ever done vid card reviews before? Get the boards to him and get him the time in his schedule to get it done. It isn't a couple hour process, it is many days running benches and doing clean installs on machines over and over again.

 

[zaydq]

Can i not that Russian, and other members including myself aren't bashing Nvidia or their products and simply promoting a product that is being shunned? Its a competetive marketplace and the controversy over all this is everyones ape shit over an alright card when theres another option. Takes two to flame war... or more.

 

[VulgarDisplay]

You ever done vid card reviews before? Get the boards to him and get him the time in his schedule to get it done. It isn't a couple hour process, it is many days running benches and doing clean installs on machines over and over again.

It almost sounds like a job. Oh, that's right it is his job.

We are all aware that doing these reviews is very time consuming. If they took some more time though their readership would benefit greatly, and more people would be interested in their reviews.

It's a pretty sad state of affairs when Tom's Hardware has the best GPU review for a new release.

 

[BenSkywalker]

Can i not that Russian, and other members including myself aren't bashing Nvidia or their products and simply promoting a product that is being shunned?

The issue is that people are expecting every site to change their policy to make the 7950 look better then what AMD decided to ship. The fault isn't with AMD for not shipping the 7950 at ~1GHZ to start with and then having a 7930 for bad binned chips, no, the fault is Ryan et al for not spending all of their time doing everything they can to make the 7950 look better.

Every generation we have tons of overclocked boards using all sorts of different chips from many different OEMs. Most of them never get tested at AT, it is only now an issue because it is the 7950. That is absurd.

It almost sounds like a job. Oh, that's right it is his job.

His job is to test GPUs. His job title doesn't say 'every off spec OEM limited edition custom built board'. You know how the chip overclocks, how it scales, if you can't put the two together to figure out how the aftermarket parts are going to perform then you probably shouldn't be installing a video card.

 

[VulgarDisplay]

His job is to test GPUs. His job title doesn't say 'every off spec OEM limited edition custom built board'. You know how the chip overclocks, how it scales, if you can't put the two together to figure out how the aftermarket parts are going to perform then you probably shouldn't be installing a video card.

If that's the case than how can you even be posting anything that is pro 660ti at it's current price point.

 

[zaydq]

The issue is that people are expecting every site to change their policy to make the 7950 look better then what AMD decided to ship. The fault isn't with AMD for not shipping the 7950 at ~1GHZ to start with and then having a 7930 for bad binned chips, no, the fault is Ryan et al for not spending all of their time doing everything they can to make the 7950 look better.

Every generation we have tons of overclocked boards using all sorts of different chips from many different OEMs. Most of them never get tested at AT, it is only now an issue because it is the 7950. That is absurd.

I see your point. But it couldn't hurt for a 7950 revisit. Prices have dropped drastically, and as such, it should be noted. People google reviews and out dated reviews don't do any justice for this card or any card.

If you're not an AMD fan or vice versa thats fine, but if one company goes, the other company will surely take advantage of having near complete marketshare.

7950s were shit at their price, but now that they're priced accordingly, it shouldn't be overlooked. The HD 7850,7950 and 7970 (screw the 7870) are all worthy contenders at their respective prices, and they should be accepted as a viable graphic solution.

In my day i've had an MX 440, an FX 5200, an X800XL, X850XT, 6600GT, 8800GTS, 8800GTX, HD 7850, HD 7950. I'm no fan boy, i buy what makes sense for my needs. I jump and praise camps all the time, theres a point to be made and its being fought so badly that this has all become a flame war.

The 660TI is not worth its msrp of $300. The free BL2 promo is nice, but won't be the sole reason this sku's pricing drops, Nvidia maybe pays $15 per BL2 product key, its a win win for both Nvidia and for the game designer studio. People need options, not just "buy a 660TI".

 

[BenSkywalker]

If that's the case than how can you even be posting anything that is pro 660ti at it's current price point.

What have I posted that is pro 660Ti? I'm not saying I couldn't post things that are, just as I could post a lot of pros for the 7950. The point is we have a bunch of posters taking the stance of rabid frothing at the mouth loyalists that hate reality and are bashing people who did a good job for reporting on precisely that.

 

[VulgarDisplay]

The thing is all the people who are touting the 7950 as the better card by far at $300-330 are the same people who would be saying that the 660ti is the better card if it was $250-275.

The 660Ti is crap at it's current price point especially when you consider the any future usage of compute in upcoming games. Anyone buying it at this price is doing themselves a major disservice, and the people who are pro 7950 are just stating facts based upon current prices.

 

[Fire&Blood]

A lot has changed since the 7950 hit the market, roughly 6 months ago. It has undergone price cuts and some serious hardware and software changes. With the latest driver and the 7970 PCB at 320$ AR, I would argue it's a completely different product. Like it's the 7960 or something.

 

[zaydq]

The thing is all the people who are touting the 7950 as the better card by far at $300-330 are the same people who would be saying that the 660ti is the better card if it was $250-275.

The 660Ti is crap at it's current price point especially when you consider the any future usage of compute in upcoming games. Anyone buying it at this price is doing themselves a major disservice, and the people who are pro 7950 are just stating facts based upon current prices.

Thanks for noticing that we've been touting for both sides throughout this thread and the other 660Ti thread.

I will also note there are users on this forum with Nvidia cards touting the same thing, its not fanboyism, its just fact and imo a service to potential buyers who are looking at their options.

 

[Siberian]

The thing is all the people who are touting the 7950 as the better card by far at $300-330 are the same people who would be saying that the 660ti is the better card if it was $250-275.

The 660Ti is crap at it's current price point especially when you consider the any future usage of compute in upcoming games. Anyone buying it at this price is doing themselves a major disservice, and the people who are pro 7950 are just stating facts based upon current prices.

The reviews say otherwise.

 

[RussianSensation]

This only became an issue when his bold predictions that the 7950 was significantly better then the 660Ti prior to launch didn't pan out in reviews.

You seriously have this conspiracy theory all figured out? It can't possibly be the case that I am asking in general for more fair representation of SKUs when new cards launch. GTX660Ti is a new change in the marketplace and there have been many new 7950 SKUs to launch since then. That may warrant a 7950 round-up. Even if the 660Ti didn't launch, I think AT's videocard reviews would strongly benefit from follow-ups of SKUs.

Computerbase has done it with 660Ti, 670/680. They also re-visited HD7950 when new significant SKUs arrived despite 7950 launching January 31st. They tested some new 7950 SKU as recently as May 17, 2012. When 660Ti is positioned as a close competitor to the 7950, why is it unreasonable to have a round-up of 7950s now that the market has changed? That's biased somehow?

Tom's Hardware tested 6 GTX670 SKUs: http://www.tomshardware.com/reviews/geforce-gtx-670-test-review,3217.html

I want to read that at AT's, which is why I come here and not to Tom's Forums. Do you know how many people on our board STILL ask what the best 670 to buy is? Maybe because they have no idea since we don't have a Mini-GTX670 Round-Up Review.

Just today someone asked "GTX 670: ASUS v. EVGA v. MSI": http://forums.anandtech.com/showthread.php?t=2265092

This shouldn't be the case if there was a proper GTX670 Mini-Roundup that addressed this for our readers. The thing is it was done at AT in the past with 6800GS, 7800GT, X800GTO.

These Mini-GPU Roundups were extremely useful for many people. Today, they aren't there anymore, not in the same form. What happens is the reviews include 2-3 after market cards during initial launch and when major changes happen in the market-place it's not revisited.

The same would be great for GTX670, 680, 7970. The reason I finally spoke up about this is because it's especially a glaring omission during $300 660Ti / 660Ti 3GB $340 vs. 7950 situation since there are at least 2-3 $330 7950 new SKUs that haven't been tested on AT and are very close competitors to the 660Ti. You being strongly opposed to such a round-up is a lot more questionable imo than Ryan considering doing more of these types of VGA Round-ups for BOTH NV and AMD that can actually help many of us here buy our next card!! More information is better and it doesn't hurt to ask if there is a demand from many readers.

This isn't about AMD vs. NV as you keep wanting to paint it.

Remember Retail 6600GT Round-up with 11 SKUs.

Why can't we have that back for our readers for 660Ti/7950/7970/680, etc.? GTX660Ti and HD7950 Mini-Round-Ups may be a great time to start bringing back the Mini-GPU Roundup tradition at AT

 

[VulgarDisplay]

The reviews say otherwise.

Anyone who can actually read into the reviews and knows anything about the current GPU landscape disagrees with you.