Best Graphics Cards For The Money: May 2011

[logainofhades]

Deactivated. This item is currently out of stock and it may or may not be restocked.

Yea I am glad I picked mine up when I did.

 

[akula2]

Could anyone tell when this card would get released?

Sapphire 6950 2GB Toxic Edition.

I'm in two minds whether to buy a 6950 or GTX 570 ($275 vs $350 ave. price). I know GTX has advantage for 3D etc but baring that I wonder if 6950 Toxic edition would match the performance of the latter? Thanks

 

[akula2]

@The Saphire 5850 Xtreme 1gb for $139-$150 blows the 560ti out of the water in terms of Price AND Performance

Yep, I do use a few Sapphire 5850 Toxic cards and they are awesome plus a great value to the investment and it runs Games awesome too.

 

[Guest]

Tom's Hardware doesn't seem to take anything but the reference models of cards in to account in making its recommendations, and that's doing a tremendous disservice to some of the card and chip manufacturers. In particular, many cards from second party manufacturers have superior cooling, a factory overclock and other optimizations that makes them perform far better than the reference model while costing only marginally more or, in some cases, not costing more at all. These overclocked models are also outselling the reference, so that the "typical" card based on certain GPUs are the overclocked models.

There is a tremendous number of cards on the market right now with a variety of timings and cooling methods. I can understand that Tom's would want to keep their recommendations generic, but feel they're doing their readers a disservice by failing to stress that fact as much as they should, or keep up with current trends in terms of what the typical clock speed / performance is for particular GPUs.

The best example of this right now is the nvidia gtx 560ti. The reference speeds for this card are 822mhz core clock, 1645mhz shader clock and 4000mhz memory clock, yet overclocked versions are selling in larger numbers, making them the "typical" 560Ti. The speeds for these OC versions can go as high as 1000mhz core clock, 2000mhz shader clock and 4500mhz memory clock. These higher clock speeds cause overclocked 560ti cards to compare much more competitively against AMD cards in the price range.

Tom's Hardware should try to incorporate OCed cards in to their recommendations by selecting models which better represent what is actually selling. eg. pick something like the MSI TwinFrozr 2 GTX 560ti or another similar card as representative of current consumer trends and then make recommendations about "overclocked GTX 560ti cards" working from that basis.

 

[gm0n3y]

FYI, NCIX.com has a sale right now on the 6870 for $164 (after $65 instant rebate and $20 MIR):

http://www.ncix.com/products/index.php?sku=56262&vpn=HD687AZNFC&manufacture=XFX&promoid=1026

 

[AMD_pitbull]

[citation][nom]Draxtier[/nom]Tom's Hardware doesn't seem to take anything but the reference models of cards in to account in making its recommendations, and that's doing a tremendous disservice to some of the card and chip manufacturers...representative of current consumer trends and then make recommendations about "overclocked GTX 560ti cards" working from that basis.[/citation]

Good to see that you have a comprehension of how many variants of each card there are, and the varying price points of those variants. If he was to take into account all the variants, people would complain about taking into account all the price points. Then you'd be looking at cards sometimes crossing 2 or 3 different prices points, all because the variants are THAT wide with regards to pricing. You get a good idea where the variant sits with regards to it's price. You find an OC'd one that's the same price? Cool. Go for it. But you're asking him to do 10 times the work for even less appreciation? Quick sum-up on that request. No, and I don't blame him.

 

[alikum]

[citation][nom]AMD_pitbull[/nom]Good to see that you have a comprehension of how many variants of each card there are, and the varying price points of those variants. If he was to take into account all the variants, people would complain about taking into account all the price points. Then you'd be looking at cards sometimes crossing 2 or 3 different prices points, all because the variants are THAT wide with regards to pricing. You get a good idea where the variant sits with regards to it's price. You find an OC'd one that's the same price? Cool. Go for it. But you're asking him to do 10 times the work for even less appreciation? Quick sum-up on that request. No, and I don't blame him.[/citation]
I strongly agree with you on this one pitbull

 

[Guest]

The problem is, recommendations about GPUs based solely on the reference models are

both useless and misleading in situations where reference models aren't representative of what's actually being bought and sold during the period for which the recommendations are intended. This is an article that's supposed to recommend video cards for May of 2011, so it should actually represent the best value in May of 2011, based on current trends in both price and technology. Instead the recommendations appear to be based on the current prices for technology that was popular months ago but has since been replaced in the market by cards which offer superior performance and value while still based on the same GPU.

It's entirely reasonable for Tom's to only look at a single model of any particular GPU, given the staggering variety of them out there. But what I would strongly encourage is for the author of this article to try to look at a model and price point that best represents the market for the period of the recommendation, rather than the reference card, which might be half a year or more out of date. If keeping up with trends in this way is beyond the scope of this series of articles, I'd question whether a new set of recommendations every month is justified.

 

[ilysaml]

i love Radeon HD6970

 

[AMD_pitbull]

[citation][nom]Draxtier2[/nom]The problem is, recommendations about GPUs based solely on the reference models are both useless and misleading in situations where reference models aren't representative of what's actually being bought and sold during the period for which the recommendations are intended. This is an article that's supposed to recommend video cards for May of 2011, so it should actually represent the best value in May of 2011, based on current trends in both price and technology. Instead the recommendations appear to be based on the current prices for technology that was popular months ago but has since been replaced in the market by cards which offer superior performance and value while still based on the same GPU.It's entirely reasonable for Tom's to only look at a single model of any particular GPU, given the staggering variety of them out there. But what I would strongly encourage is for the author of this article to try to look at a model and price point that best represents the market for the period of the recommendation, rather than the reference card, which might be half a year or more out of date. If keeping up with trends in this way is beyond the scope of this series of articles, I'd question whether a new set of recommendations every month is justified.[/citation]
Tell you what, after he's done checking the price and performance of every single modle of every single card, how about he does all the conversion rates too, checking every country to see if the prices vary and by how much, so he can make a guide for everyone? Sound ridiculous? It's only one step further than what you're asking.

 

[Guest]

What I'm proposing would involve getting information from the more popular retailers, or even just one of them (newegg, amazon, bestbuy, tigerdirect) to determine which models of any particular GPU are selling and basing recommendations from that.

Tom's Hardware is a leading online resource and their recommendations bring a lot of traffic to retailers. This creates a commercial incentive for retailers to be as nice and helpful to Tom's as they can be. Consequently, I imagine people at places like newegg.com or amazon.com could provide this sort of anonymous consumer information to Tom's Hardware, if it was requested. ie, which model of cards based on the Radeon 560ti GPU sold the most in the month of April? Retailers already track this information as part of their general business practices.

So... what I am proposing is that Don Woligroski contact the retailers I've mentioned, and any others that might prove helpful, and explore what sort of help they can provide to make the recommendations more reflective of actual trends. As mentioned, the retailers have a strong financial incentive to be as helpful as they can be, and they'd have this sort of information already available.

So... nothing ridiculous about it, AMD_Pitbull.

(note: I'm having trouble registering with Tom's. Until that's fixed, I'll be an incremental Draxtier#)

 

[masterofevil22]

Wow, ATI almost gets a clean sweep this month..

Too bad there are no games to push pc's to their limits these days.

Oh well, guess we'll just have to wait on the next ps and xbox to up the development anti until we can actually use any of this crap for anything except slightly cleaner versions of console games

 

[AMD_pitbull]

[citation][nom]Draxtier3[/nom]What I'm proposing would involve getting information from the more popular retailers, or even just one of them (newegg, amazon, bestbuy, tigerdirect) to determine which models of any particular GPU are selling and basing recommendations from that.Tom's Hardware is a leading online resource and their recommendations bring a lot of traffic to retailers. This creates a commercial incentive for retailers to be as nice and helpful to Tom's as they can be. Consequently, I imagine people at places like newegg.com or amazon.com could provide this sort of anonymous consumer information to Tom's Hardware, if it was requested. ie, which model of cards based on the Radeon 560ti GPU sold the most in the month of April? Retailers already track this information as part of their general business practices.So... what I am proposing is that Don Woligroski contact the retailers I've mentioned, and any others that might prove helpful, and explore what sort of help they can provide to make the recommendations more reflective of actual trends. As mentioned, the retailers have a strong financial incentive to be as helpful as they can be, and they'd have this sort of information already available.So... nothing ridiculous about it, AMD_Pitbull.(note: I'm having trouble registering with Tom's. Until that's fixed, I'll be an incremental Draxtier#)[/citation]
See, herein lies the problem with gathering the information from those retailers, once again, as previously stated, you're taking variants which change in price daily, and comparing a reference to a variant that could be over-clocked, different cooling, etc. which would make it more fair to look up the variants and perfomance of the other cards. Considering the variants come out a lot more than just monthly, and that "best sellers" don't necesarrily represent the top performance, it makes it pretty well pointless to go off that information. Say, for instance, one version of the 550 out-performs another by 10%, but, the lower performer has a lower clock and only $10 difference. If the lower clocked one sells better, and the difference is 140 compared to 150, the 150 would be the better deal with regards to the performance increase, but, wouldn't be considered due to the fact that it's not selling as well. He's going off the cards they have, the data they collected, and the prices at the avg going rate. You wanna do your own research, cool, go for it. Every enthusiast should, and probably will. He's giving you a basic guide showing where the cards lie at each level, and how well they perform.

 

[PCGOD]

for people looking for a good card at around 110$, forget the 5770. the sapphire 5830 is 110 on newegg now and is a better card than the 5770. ordered one for a freidn today, cant wait to try it out.

 

[Guest]

I'm talking about making an attempt to pick a middle of the field, average card in terms of price / performance for the period, Pitbull, instead of using the reference. You keep repeating that it's impossible to pick one card because there's an overwhelming number out there, when in fact, they're already basing recommendations off of a single, arbitrarily chosen card: the reference model. It's just a really bad representation of certain GPUs at this point.

Have a look here http://www.tomshardware.com/reviews/geforce-gtx-560-ti-roundup-asus-engtx560-graphics-card-overclocking,2858.html then tell me they don't have the cards or the data to give better recommendations in the monthly articles. There are five OCed 560ti nvidia cards. Check the prices on newegg, construct an aggregate of the price / performance ratio that considers all five OCed cards and average them, then make a recommendation based on "Overclocked gtx 560ti cards". If I had access to Tom's own internal data, and cross referenced that against prices on newegg and amazon et al. it would take me five minutes. Unfortunately, in comparing the five OCed models to one another, Tom's didn't compare them to any sort of reference, although the Jetway card was supposed to fill that role in the test and Tom's can't really be blamed. Still though, it's unfortunate.

Some attempt, ANY attempt, should be made to incorporate variant models like these when variant models dominate the market. There are nearly limitless ways to attempt to do this, all with varying degrees of accuracy and difficulty. I outlined one way of attempting it in my previous post and a second way in this post.

The crux of your rebuttal, Pitbull, seems to be that incorporating variant cards is A) a more difficult method than just using the reference and B) an imperfect method of gauging the market. I concede to both those points. However making an effort to incorporate variant cards is also A) not nearly as difficult as your hyperbolic rhetoric asserts and B) vastly more useful and helpful to the consumer than recommendations that only consider the reference model, despite any imperfections it may entail.

Looking at newegg.com's available stock of nvidia 560Ti cards I find that they carry 24 models. Of these 8 have reference or near reference core clock speeds (820~835mhz) and a total of 83 reviews, 4 have moderate factory overclocks (850~880mhz) and a total of 293 reviews between them, 9 have high overclocks (900mhz) and a total of 198 reviews, the final 3 have extreme overclocks (950mhz) and 45 reviews between them. If you look at the number of reviews and take them as an indicator of popularity and sales, it's possible to estimate that the reference model represents about 13%, moderately overclocked models make up 47%, highly overclocked models make up about 32% and extremely overclocked models of the gtx560 make up 7% of the market. Obviously these numbers are inaccurate, but they illustrate the problem of only accounting for reference models.

In contrast, models of the Radeon 6950 all conform much more closely to the reference model in terms of clock speeds. Consequently, when you compare the costerformance ratio of just the reference cards, Radeon comes out the clear winner, but that's failing to account for the majority of nvidia cards. It's an obvious bias in the recommendations that's favoring one brand over the other as a consequence of Tom's very poor methodology in making these comparisons.

If you choose to respond again, Pitbull, please address the itemized points I've made and perhaps contribute something to the discussion that isn't purely a matter of your unsupported opinion and speculation.

 

[AMD_pitbull]

[citation][nom]Draxtier4[/nom]I'm talking about making an attempt to pick a middle of the field, average card in terms of price / performance for the period, Pitbull, instead of using the reference. You keep repeating that it's impossible to pick one card because there's an overwhelming number out there, when in fact, they're already basing recommendations off of a single, arbitrarily chosen card: the reference model. It's just a really bad representation of certain GPUs at this point.Have a look here http://www.tomshardware.com/review [...] ,2858.html then tell me they don't have the cards or the data to give better recommendations in the monthly articles. There are five OCed 560ti nvidia cards. Check the prices on newegg, construct an aggregate of the price / performance ratio that considers all five OCed cards and average them, then make a recommendation based on "Overclocked gtx 560ti cards". If I had access to Tom's own internal data, and cross referenced that against prices on newegg and amazon et al. it would take me five minutes. Unfortunately, in comparing the five OCed models to one another, Tom's didn't compare them to any sort of reference, although the Jetway card was supposed to fill that role in the test and Tom's can't really be blamed. Still though, it's unfortunate.Some attempt, ANY attempt, should be made to incorporate variant models like these when variant models dominate the market. There are nearly limitless ways to attempt to do this, all with varying degrees of accuracy and difficulty. I outlined one way of attempting it in my previous post and a second way in this post.The crux of your rebuttal, Pitbull, seems to be that incorporating variant cards is A) a more difficult method than just using the reference and B) an imperfect method of gauging the market. I concede to both those points. However making an effort to incorporate variant cards is also A) not nearly as difficult as your hyperbolic rhetoric asserts and B) vastly more useful and helpful to the consumer than recommendations that only consider the reference model, despite any imperfections it may entail.Looking at newegg.com's available stock of nvidia 560Ti cards I find that they carry 24 models. Of these 8 have reference or near reference core clock speeds (820~835mhz) and a total of 83 reviews, 4 have moderate factory overclocks (850~880mhz) and a total of 293 reviews between them, 9 have high overclocks (900mhz) and a total of 198 reviews, the final 3 have extreme overclocks (950mhz) and 45 reviews between them. If you look at the number of reviews and take them as an indicator of popularity and sales, it's possible to estimate that the reference model represents about 13%, moderately overclocked models make up 47%, highly overclocked models make up about 32% and extremely overclocked models of the gtx560 make up 7% of the market. Obviously these numbers are inaccurate, but they illustrate the problem of only accounting for reference models.In contrast, models of the Radeon 6950 all conform much more closely to the reference model in terms of clock speeds. Consequently, when you compare the costerformance ratio of just the reference cards, Radeon comes out the clear winner, but that's failing to account for the majority of nvidia cards. It's an obvious bias in the recommendations that's favoring one brand over the other as a consequence of Tom's very poor methodology in making these comparisons.If you choose to respond again, Pitbull, please address the itemized points I've made and perhaps contribute something to the discussion that isn't purely a matter of your unsupported opinion and speculation.[/citation]

I've already addressed every "itemized point" you've made, to be quite frank. You keep pushing the variants and put down the reference cards. You use the 560ti as an example, one that proves MY point, actually. You say there's 5 OC'd models of this one alone. That's not including reference or low power, or water cooled, etc. 24 in total I believe you stated. You say I'm ignoring things, but, how would you compare these cards and which one to include? Is the reference better price performance than the OC'd model? Which level of OC is the best price/performance? You're talking about him having to put almost as much work into this one card as per the entire article. Taking a best seller off a list blindly and giving his recommendation of it defeats the purpose to begin with, once again, bringing us back to him having to put more work in, for less effort.

Your statement with regards to AMD being the clear winner when using reference cards makes you sound biased towards Nvidia. Not sure if that's the case, but, thought I'd let you know.

I'm not saying it's impossible, nor have I ever implied that. If you read my comments, you'd see that I said, as I stated from the beginning, that it's simply more work and making a quick reference guide for people is not his only job there. I said that taking into account the variants would mean you have to take into account the price points they represent as well, and, as menial a task as you want to make it seem, when using those, you have to take into account the fact that the price of those can change on a daily basis and don't have the same MSRP that the reference cards are going to have due to different coolers, clocks, etc. Saying that it's really bad representation of certain GPU's at this point? So, he should take the variants into consideration due to a couple cards not being done right at the reference level? Sounds like a marketing problem at a corporate level, not Tom's problem. As for me saying it's "an imperfect method of gauging the market", I was stating that sites can vary and the best selling card might not be the best for the money.

The amount of work he'd have to throw into the dozens of extra cards he'd have to take into consideration is not worth the effort. You want to include the OC'd, twin cooled, water cooled, etc. Make a comment about deals after. This is putting a basic idea of where the cards sit. Yes, taking only reference cards into account will more than likely not allow certain cards to really show off what they can do. That's one of the sacrifices. They have testing done with OC'ing, scaling, etc. with reference cards, all of which can make a stronger point for those who want a reference design. If you don't, or you disagree with the AMD/Nvidia choice, as you clearly do, look into whatever you'd prefer. Simple.

Basically, once again, the simple fact of the matter is that if you don't like how he does these, don't use them. The levels he puts the cards at is basically where they should be with regards to pricing and performance. You can argue that it's "unfair" or "not properly representing the consumers" or whatever else you want, but, in the end, it's the most fair way to deal with the reviews. I really do hope you're not taking this personal, Drax, cuz it's kinda coming across that way. Especially when you say my comments are "unsupported opinion and speculation" when comparing to your comments which are nothing more than ideas. Funny, considering that my comments have already were supported opinions by even the author of the article, other readers, and yourself in this comment. I'm glad you want to try to contribute ideas, but, attempting to make yourself seem better by putting vague personal insults into a discussion make you seem a lot less credible and rather frustrated.

Sorry for the long post this time around, it's tired, and reading a comment that's coming out personal from the other end tends to make me not want to respond as much. If you're getting upset over an article, usually time to move on. But, I look forward to your next post, Drax, as your unsupported opinion and speculation on how Cleeve should be doing his job and how he's not doing it right, and how your ideas are right when his are wrong, is giving me something to read until Cleeve's next review. Appreciate it

 

[Draxtier]

Well, I'm pleased you're enjoying yourself, Pitbull!
Personally, I'm finding this exercise tedious.

To clarify what's been going on here: I've gone out and found the info, Pitbull. It didn't take me too long and provided me with a firm position from which to make my points. I read this review, I looked up the cards, I checked the prices, I crunched the numbers. This isn't my job at all, but I put in the effort because I think it's worthwhile.

What you've consistently done is to speculate while offering no tangible information beyond your own opinions and perceptions about what is easy and what is hard, what is reasonable and what is unreasonable. Absolutely anyone can do that, and opinions can vary wildly. I'd just like to know where your opinions come from. For all I know, you have shaken a magic 8 ball and taken your opinions from that. Is that the case? Take some time, go back and review what's been stated on both sides. You're right in thinking that I am getting frustrated. While I am trying to form persuasive arguments based upon a foundation of facts, backed up by a bit of effort and research, you are providing opinions. Disagreeing with your opinions, or anyone's opinions for that matter, is a pointless exercise especially on the internet. What I would like to do rather than address your opinions is to address the facts your opinions are based upon. If you could provide some facts that substantiate your opinions, that would be immensely helpful.

Incidentally, I have a Radeon 6850 card in my PC with an AMD Phenom II CPU. Before this, I had another Radeon card and an Intel CPU. I have no personal bias for nvidia or against AMD. I was pointing out a bias in the recommendations Tom makes due to a flaw in the methodology they use in analyzing a card's value. My statement with regards to AMD being the clear winner comes from reading the review, which states: "At $235, the new GeForce GTX 560 Ti is an impressive card, and from a price/performance perspective, it offers a very similar value proposition to the card it's replacing, Nvidia's GeForce GTX 470. But at this price, it is far too close to the more powerful Radeon HD 6950 1 GB." That quote is only true while you are comparing the reference cards. If you compare a 1gb Radeon 6950 to a 'middle of the pack' GTX 560ti (I'd go with the MSI Frozer II at the moment), the 560 consistently performs comparably or slightly better on benchmarks, (winning slightly more than it loses, given a range of applications / games) and it's cheaper. This helps to illustrate the difference between opinions, misconceptions and bias. Just going by your observation about my statements makes it sound like you never actually read the review we've been discussing. Not sure if that's the case, but, thought I'd let you know.

Here's an analogy for you that may help clear up your confusion: In school we all learned to divide. First we learn to divide whole numbers and leave the remainder separate. Then we learn to divide and account for fractions, to divide and use a decimal and to divide and round. Those are four different methods of dividing, they entail four methodologies, each with inherent difficulties and inaccuracies. If you take the equation 4/2 and apply these different methods you get the same answer: 4/2=2 But if you take the equation 5/2 and apply them you get four different answers which are 5/2=2 (dropped remainder), 5/2=2½ (account for fraction), 5/2=2.5 (account for decimal) and 5/2=3 (rounded up) respectively. What I am contending, to use that analogy, is that in writing his recommendations I believe the author is doing the equivalent of dividing and dropping the remainder because it's easiest, this gives the result 5/2=2. I contend that this method is inaccurate and have provided evidence to illustrate the inaccuracy, the problems this creates and recommend an alternative (see previous posts). To continue the analogy, it's like I'm suggesting they account for fractions or use decimals. You have exhaustively argued that accounting for fractions is too hard, or that dropping fractions is good enough, or that any alternative would be equally inaccurate and finally that if I don't like it I should go away. These are your opinion. They are unsupported. You have not provided any sort of evidence to substantiate your opinions. I disagree with your opinions and you with mine, and that's fine. My point remains though, and it is intended more for the author. I imagine he's quite capable of speaking for himself, and knows better than either of us exactly how much work is involved in the review, and how difficult changes would be.

A few points:
Pitbull: "As for me saying it's "an imperfect method of gauging the market", I was stating that sites can vary and the best selling card might not be the best for the money. "
-- The same fluctuations in price will occur with all cards, including the reference. We can discount price fluctuations as something which occurs with all cards. This leaves only the question of whether the reference or an alternative, such as the best selling card, is a better representation of a GPU's value. I establish why a reference card is a bad choice in certain situations in a previous post. Review it for details.

Pitbull: "The amount of work he'd have to throw into the dozens of extra cards he'd have to take into consideration is not worth the effort."
-- This is your opinion. No basis for this opinion has been offered, ergo it is speculation. The author may agree with you, but then again they might not. You shouldn't speak for him. The link I provided in a previous post shows that Tom's Hardware has the information for 5 OCed versions of that specific exacmple readily available, making it a relatively simple matter to incorporate more accurate information about the 'middle of the road' gtx 560ti card. As luck would have it, one of the cards reviewed in that article IS the most popular model for sale on newegg, going by the number of reviews it has. So, at least in the case of the gtx 560ti, Tom's Hardware has everything they need to make more accurate recommendations that account for OCed cards.

Pitbull: "You use the 560ti as an example, one that proves MY point, actually. You say there's 5 OC'd models of this one alone. That's not including reference or low power, or water cooled, etc. 24 in total I believe you stated. You say I'm ignoring things, but,"
-- Ignoring things or completely missing them. Go back and read my posts again, you'll find that this has been addressed. In establishing my point, I looked at all 24 versions of the 560ti for sale on newegg. Reference models accounted for roughly 13% of sales, going by the number of reviews. If you don't understand why looking at 13% of cards and discounting 87% of them is a poor method for determining the value of a GPU, that is a problem with your grasp of sampling and statistics and resolving that problem is beyond the scope of this discussion.

Pitbull: "how would you compare these cards and which one to include?"
-- Again, go back and review earlier posts. This has been addressed in detail. The gist of it is; There's no perfect or ideal way to account for OCed cards but any attempt to account for them is better than no attempt at all.

Pitbull: "Basically, once again, the simple fact of the matter is that if you don't like how he does these, don't use them."
-- We appear to have a fundamentally different understanding of what the word 'fact' means.

Pitbull: "I really do hope you're not taking this personal, Drax, cuz it's kinda coming across that way. Especially when you say my comments are "unsupported opinion and speculation" when comparing to your comments which are nothing more than ideas."
-- I am not upset, nor taking this personally, though I do find myself regretting having engaged in the discussion to this degree. I should know better than to argue on the internet. Now that I'm in it though, I feel a certain obligation to see it through. Frankly, this discussion is beginning to remind me of trying to teach calculus. No matter how many ways I find to explain and clarify a concept, there will be people who don't get it. Hopefully this time I've expressed my points in such a way that further discussion won't be needed. My statement that your points are unsupported opinion is, as far as I can tell, accurate. It wasn't meant as an insult and you should not take it personally. I'm accustomed to expressing criticisms in clear language. In this case I mean to point out the difference between conclusions supported by evidence and unsubstantiated opinion. This concept is critical in any expository writing and something generally covered in school when we really start learning how to compose essays. I have tried to base all of my points upon evidence of some sort. You have provided none.

I'll note again that it's speculation on my part to say that Tom's doesn't account for variant (overclocked) cards. Maybe they do and it's just not apparent in the review. If that's the case, I can go back to being a passive reader until I have something else to contribute. If it's not the case, I'd really like to have the matter addressed. At the very least, a note should be added to the first page to clarify that the article doesn't take any OCed cards in to account in making its recommendations, and readers should bear this in mind, especially in cases where OCed versions of a GPU are more prevalent than cards at reference speeds. Of course, making an effort to account for OCed cards, at least in general, would be preferred.

Finally, I also really value and appreciate the service Tom's Hardware provides and I've been reading their articles and reviews for years without ever having posted a single comment. I never felt any need to. It's good stuff and useful and has helped me a lot over the years. I like this review too, but I believe could be better. I see problems with it I would prefer were corrected, or at least addressed. Providing constructive criticism is actually a good thing and in general, we'd all be better off if it was better received. In situations like this one, pointing out a perceived flaw is better than choosing to ignore it.

And on the bright side, the website software decided to let me register at last.

 

[AMD_pitbull]

Well, Drax, glad you could finally register and not make it up to 100+. I'm sorry to hear you feel exhausted restating your point again, but, I do have to say your analogy is rather inaccurate this time around. I've understood your point to begin with, and our argument is quite simple: You think he should include variants, I think it's more trouble than it's worth. The fact that there's more variants out there than reference doesn't make it relevant to include them, nor does it make it right. But, to use an analogy along the basis you request, against, I'll resubmit my unsupported opinion and speculation in these terms, as you think your opinions on it being important to include variants and it not being more work seems to be the fact you're pressing: The article says 1x10000=10000 is quite sufficient, where as you're pushing 100x100=2500x4=250x40=10000x1=5x5x5x5x16 etc. and which method is the best to reach the end result. It's your opinion that factoring all the variant cards and factoring the different cards at their price points isn't an issue. You've pushed that, many times. I'll stop posting as you like to keep saying that all I do is "speculate" that doing the extra work is more work. So, I wish you the best in your next "fact" proving mission and look forward to the next review.

 

[Draxtier]

Thanks for the welcome!

Neither of us know exactly what research the author does, what trouble he goes to or how much trouble it would be to make changes in that methodology. We're both speculating, and arguing that point is a waste of both my time and yours, Pitbull. Instead of continuing to do so, it will be both easier and more productive to wait for the author to address the issue I've raised at his convenience.

As for the analogy, it was actually rather good and I'm reasonably proud of it. I'd have delved in to concepts taught in a basic statistics class at the college level, which would have better represented the issue, but I fear that would only confuse you more.

 

[osiguy]

I have the Sapphire version of the 6570. For the price it's a great little card.

I just wish i could find an answer on how to get the control center software to work in MS 2008 r2.
I like many people have access to 2008 r2 and are using it as a workstation because it has some attributes that make it better than win7 and we can use it for server purposes as well. But for some reason ATI hasn't put out a ccc driver that will work with 2008 r2. People on the forums give pat smart a** answers like ...because it's a server and not supported.... duh

Well Nvidia cards run on server 2008 r2 just fine, what's wrong with ATI that's what I want to know.

 

[the_brute]

Please read the article first people. Yes this is news on statistics, statistics and be read anyway you want. Personally these people work on this every day; and its fun to see their speculations on what new gqus can do before testing. So far they are damn close.

He stated all the cards he recommended and gave further info on why that card was chosen. There are how many cards on that list to look at then searching the internet for the lowest price? Or should you look at the current MSRP and hint that many sites may have it lower.

Once again this is a very good article, and I hope nVidia will lower prices and make stiffer competition per price point.
Ill be back next month to read your next best GPU article, keep these coming.

 

[the_brute]

@Draxtier
The very first 2 paragraphs of this article.

[citation]Detailed graphics card specifications and reviews are great—that is, if you have the time to do the research. But at the end of the day, what a gamer needs is the best graphics card within a certain budget.

So, if you don’t have the time to research the benchmarks, or if you don’t feel confident enough in your ability to pick the right card, then fear not. We at Tom’s Hardware have come to your aid with a simple list of the best gaming cards offered for the money."[/citation]

 

[Draxtier]

@Draxtier
The very first 2 paragraphs of this article.

I read the article, Brute, including those paragraphs. I'm not sure why you're directing me to them though.

 

[Guest]

Okay, I always check the GPU list, but I am still confused about Radeon HD 5670. Is it DDR5 or DDR3 type is in the list? Both the cards are available.

 

[Guest]

What Draxtier has been trying to say to exhaustion is that some effort is better than none. And clearly reference cards do not dominate the market by a large margin. This is producing results that are becoming way too out of sync with current market offerings.

Necessarily checking every single possibility out there is out of the question. However the article proposes to address "Best Buys" for a given month that invariably can result on many occasions on bad advise. Some better methodology must be devised. This series is not going down the hill very fast with already too many examples of "Best Buys" that simply aren't when you actually go shopping for the real cards. Some manner of incorporating vendor solutions and prices must be thought about. Failing that I suggest, the article changes the tone from "Best Buy" to something else less elated.