AMD Radeon 6990+6970 CrossFireX / "TriFire" Review @ [H]

I get that it is a value based comparison and I AGREE with it. I'm not trying to discredit the basis at all, I'm just saying that it's 3 vs 2 from the same generation which isn't fair once you get out of a price based comparison, that's all. I don't see how this is bothering any of you since I'm not even targeting price comparison with it, I even said I agree with it from a price comparison standpoint, but I don't from any other view... I don't see why something this simple is being shot way out of proportions. People agree and disagree based on different conditions all the time, you shouldn't try to stamp out anything you don't want to hear.
 
Last edited:
I also get the point of the article as a value based comparison, in the same vein, and without beating a dead horse, or attempting to poke the already annoyed Bennett-Lion, I would have liked to see the 3 nvidia gpu $1,000 setup (570's) included in the review. Next, I'll say the data sets provided, are as always, superb - but the color commentary serves only to significantly diminish the appearance of objectivity - which traditionally has seemed to be highly valued here.
 
I rather just have loads of 6950s with unlocked shaders and OC'ed with a Bulldozer or 2600K replacement.
 
Last edited:
I also get the point of the article as a value based comparison, in the same vein, and without beating a dead horse, or attempting to poke the already annoyed Bennett-Lion, I would have liked to see the 3 nvidia gpu $1,000 setup (570's) included in the review. Next, I'll say the data sets provided, are as always, superb - but the color commentary serves only to significantly diminish the appearance of objectivity - which traditionally has seemed to be highly valued here.

What is so hard to understand? At eyefinity resolutions, the small framebuffer of the 570 will be a limiting factor. So 2,3,4 or ten 570's at eyefinity resolutions are all equal pointless.

Want to have a triple screen setup with lots of AA and such? 6950,6970 or gtx580 is the only sane answer. If you still do not get this, go have a look in the 570sli/580sli/6950cf/6970cf review. Look at the frame rates for the 570. Adding more of them will be even more pointless.
 
I would really love to see 3x 6950's that have been bios modded to 6970. The performance/price ratio would go way up if it worked well.


The performance would be identical to the 6970s but I find the bios modded 6950 runs about 4 degrees hotter then the stock 6950. That could be a problem if you have three 6950s sandwiched together.
It would be about $200 cheaper and give you better performance, since the 6970 runs at at 880 core and 5500 memory.
 
Nah. We don't need 6990+6970 to beat 580 SLI anymore. 6990 alone on water, OC alot higher tthen 2X6970, and is beating 580 SLI OC in benchmarks at those crazy clocks.

Look at that! :) Who needs Tri-Fire with that. :) And just to think you can't even OC the nvidia 590 at all...

2d7u6w3.jpg


http://www.overclock.net/13114302-post888.html
 
Last edited:
I get that it is a value based comparison and I AGREE with it. I'm not trying to discredit the basis at all, I'm just saying that it's 3 vs 2 from the same generation which isn't fair once you get out of a price based comparison, that's all. I don't see how this is bothering any of you since I'm not even targeting price comparison with it, I even said I agree with it from a price comparison standpoint, but I don't from any other view... I don't see why something this simple is being shot way out of proportions. People agree and disagree based on different conditions all the time, you shouldn't try to stamp out anything you don't want to hear.

I currently disagree with the premise that only a comparison of similar numbers of GPUs would be fair, and pose the question: why is it unfair to compare different numbers of GPUs? I don't see anything wrong with these setups, both use up the same number of slots, and cost the same, while one exhausts some of its spent air into the case and the other exhausts all of it, and one supports PhysX while the other does not. Why is it unfair if for all intents and purposes they each behave as a single CrossFire/SLI setup?

EDIT: To make an analogy, do you think it's unfair for Intel's quad-core processors to compete against AMD's six-core (and in the future eight-core) processors? In my opinion the answer for CPUs should be the same for GPUs, in the end one uses more units, nothing wrong with that to me.

EDIT2: D'oh! OK now I get it, and I agree with your premise, it's only fair when they're in the same price ranges, absolutely :).

I also get the point of the article as a value based comparison, in the same vein, and without beating a dead horse, or attempting to poke the already annoyed Bennett-Lion, I would have liked to see the 3 nvidia gpu $1,000 setup (570's) included in the review. Next, I'll say the data sets provided, are as always, superb - but the color commentary serves only to significantly diminish the appearance of objectivity - which traditionally has seemed to be highly valued here.

http://www.tomshardware.com/reviews/crossfire-sli-3-way-scaling,2865-3.html

You can take my interpretation with a grain of salt, but based on this review 570 TriSLI would probably get spanked. 1280MB of VRAM is not enough for 3x a GTX 480's performance. I'm so confident in this statement that I'd say [H] would be wasting its time to make that comparison. But there you go, enjoy.
 
Last edited:
I currently disagree with the premise that only a comparison of similar numbers of GPUs would be fair, and pose the question: why is it unfair to compare different numbers of GPUs? I don't see anything wrong with these setups, both use up the same number of slots, and cost the same, while one exhausts some of its spent air into the case and the other exhausts all of it, and one supports PhysX while the other does not. Why is it unfair if for all intents and purposes they each behave as a single CrossFire/SLI setup?

EDIT: To make an analogy, do you think it's unfair for Intel's quad-core processors to compete against AMD's six-core (and in the future eight-core) processors? In my opinion the answer for CPUs should be the same for GPUs, in the end one uses more units, nothing wrong with that to me.

EDIT2: D'oh! OK now I get it, and I agree with your premise, it's only fair when they're in the same price ranges, absolutely :).



http://www.tomshardware.com/reviews/crossfire-sli-3-way-scaling,2865-3.html

You can take my interpretation with a grain of salt, but based on this review 570 TriSLI would probably get spanked. 1280MB of VRAM is not enough for 3x a GTX 480's performance. I'm so confident in this statement that I'd say [H] would be wasting its time to make that comparison. But there you go, enjoy.

I don't care what pwns what if it is the same number of GPU's. That's the issue, I already accept the price point argument, but not any other.

When it comes to CPU's it is also the same for me, sure you can do a price point comparison. However, I would always rather see 4 cores vs 4 cores, rather than 6 vs 4 in addition to price point comparisons.

I would just always like to see both and not just narrow down to one thing only and run with it.
 
I don't care what pwns what if it is the same number of GPU's. That's the issue, I already accept the price point argument, but not any other.

When it comes to CPU's it is also the same for me, sure you can do a price point comparison. However, I would always rather see 4 cores vs 4 cores, rather than 6 vs 4 in addition to price point comparisons.

I would just always like to see both and not just narrow down to one thing only and run with it.

As valid as this opinion is (I agree both views are neccessary), if you wanted [H] to cover that as well you have to understand their time constraints with their testing methods.

People are going to want to look at the best possible set up(even grounds?), the majority will want to look at the price vs price set up. I think the [H] was right in doing the price vs price for their review, it seems to be getting a lot of good attention.
 
bah whatever take the review apply your own logic filter from there and decide what works for you, pointless arguing your pointless point. Much master debating going on in this thread with noobie accounts going full tilt at the windmills.
 
As valid as this opinion is (I agree both views are neccessary), if you wanted [H] to cover that as well you have to understand their time constraints with their testing methods.

People are going to want to look at the best possible set up(even grounds?), the majority will want to look at the price vs price set up. I think the [H] was right in doing the price vs price for their review, it seems to be getting a lot of good attention.

I don't want [H] to cover anything additionally, they did price comparison and I agree with their comparison. I'm not bashing the writers, the site or anyone. All I'm saying is that I like to see Tri vs Tri outside price point comparison. Nobody is obligated to do it and I have no problem with it. However, it should also be fine for me to agree with this review and site something that I also like to see too (without demanding it be done). I haven't done anything out of the ordinary, I see way too much fuss over a small matter.
 
I don't want [H] to cover anything additionally, they did price comparison and I agree with their comparison. I'm not bashing the writers, the site or anyone. All I'm saying is that I like to see Tri vs Tri outside price point comparison. Nobody is obligated to do it and I have no problem with it. However, it should also be fine for me to agree with this review and site something that I also like to see too (without demanding it be done). I haven't done anything out of the ordinary, I see way too much fuss over a small matter.

Then it isn't fair to ATI because you are comparing two products with very different price points. Just like it isn't fair to compare a 6970 vs. a GTX 560 - two different price points.

It is *always* about price except for a single case - the ultra high end, which is where quadfire and quadsli are. Every other comparison is always about price. They always have been.
 
Then it isn't fair to ATI because you are comparing two products with very different price points. Just like it isn't fair to compare a 6970 vs. a GTX 560 - two different price points.

It is *always* about price except for a single case - the ultra high end, which is where quadfire and quadsli are. Every other comparison is always about price. They always have been.

Totally agree with that. Same price point is the only good and valid comparison. Except for the extreme high-end.

And with the much better AMD 6xxx scaling, we all lknow 6970 Quad-Fire would easily beat Quad-SLI 580, since Nvidia scaling is inferior right now. And 6970 would cost ALOT less.
 
Wow AMD owns! Too many green butt hurt boobies I see :D

I'm happy about these results.

Hopefully there will be more 580's on sale now so I can get a second one for SLI cheaper :p

The way I see it is that I deemed the 580 a good board when I bought it for what I wanted to do and was always planning to go SLI when I saved up. The fact that there is now something else out there that may beat it doesn't mean that the card is any worse that it was before :p

I have nothing against AMD, except that their drivers are not as solid under Linux as the Nvidia closed source drivers are, and I am a huge Linux user.

So that being said, my shopping list for video cards usually includes "whatever Nvidia card is fastest at the moment".

IMHO, if AMD improves their performance that is really just a good thing for the end user as it drives cost improvements and more competition. I think Nvidia has recovered enough from their delayed Fermi launch at this point that no one needs to be concerned about them going under anymore. (I don't know if this was a legitimate concern to begin with, but it was voiced for a while).
 
The 580 is a good card. I just don't understand the debate of 3 GPU vs 2 GPU. This was a really well done review comparing what you can get for $1000. For most people, don't we spend on a budget? (unless you had an unlimited income, then who are you, Bill Gates?) Heck I'm running a X1900GT right now, and it's painful. I was running CF 6950 2GB when they were released, and because of financial debt I had to sell them for cash. Thank god I am able to keep my SB rig however! Can't imagine selling it for something slower.

Back on topic. Tri-SLI 570 could work but it has already been shown than even that setup hits VRAM limitations at 30" resolutions let alone dismal performance at 3x24" displays as [H] has shown. For cheap Triple Card gaming, why compromise ($289AR) image quality then there is a superior solution for an even cheaper price ($235AR), unless you have a specific brand preference but then there's no one else to blame but oneself.

So entry into Green Team for $870AR or Red Team for $705AR. Pick your poison. Pay less BUT you get MORE. OMG that's SO WRONG!!

Edit: Also you can't Tri-SLI 2GB 560TI's so that goes right out the window.
Oh, for the price of Tri-SLI 580s you can buy Tri-CFX 6970 AND Triple 24" TN LCDs.
 
Then it isn't fair to ATI because you are comparing two products with very different price points. Just like it isn't fair to compare a 6970 vs. a GTX 560 - two different price points.

It is *always* about price except for a single case - the ultra high end, which is where quadfire and quadsli are. Every other comparison is always about price. They always have been.

I'm not talking about price. It's still 3 vs 2 outside of a price point comparison, obviously the one with the higher number has a clear advantage which doesn't make this fair once you get out of comparing them by price. I know this article has nothing to do with that, but at the same time you can't turn a blind eye to details you don't like.
 
I'm not talking about price. It's still 3 vs 2 outside of a price point comparison, obviously the one with the higher number has a clear advantage which doesn't make this fair once you get out of comparing them by price. I know this article has nothing to do with that, but at the same time you can't turn a blind eye to details you don't like.

betcha I can.

it was an interesting article but well beyond my price limits of hardware. Fun to watch people play with expensive toys!
 
I'm not talking about price. It's still 3 vs 2 outside of a price point comparison, obviously the one with the higher number has a clear advantage which doesn't make this fair once you get out of comparing them by price. I know this article has nothing to do with that, but at the same time you can't turn a blind eye to details you don't like.

Out of curiousity, do you go on car forums and complain when the performance of two cars are compared against eachother that are in the same class: Ie Luxury and same price range: Ie both cost $26,000.00 only because one company offers a V6 at that price and the other company offers a V4? Do you complain you'd like them to load a custom bios onto the fuel injectors to disable 2 pistols onto the V6 so the comparison of performance is more 'fair' regardless of the fact that if a customer spent the same $26,000.00 the actual car they'd get would have 6 cyclinders? Wouldn't that just promote the adding of mis-information?

Sorry, but, most people who quad-sli or tri-sli have money. People with money buy top end-cards. They don't buy 2nd tier cards x 3. They'd rather buy top tier x2 b/c top tier cards have more ram on their GPU generally providing more performance at high resolution/high levels of AA which is the purpose of quad/tri-SLIing to begin with. The number's your requesting for tri-core versus tri-core comparison of 3x570 versus 3x6970s would be a waste of the reviewers time as those who could afford 3x570 would realize 2x580 is better performance and buy that instead.
 
I'm not talking about price. It's still 3 vs 2 outside of a price point comparison, obviously the one with the higher number has a clear advantage which doesn't make this fair once you get out of comparing them by price. I know this article has nothing to do with that, but at the same time you can't turn a blind eye to details you don't like.

The 6900 setup doesn't have a clear advantage because it has more GPUs, it has a clear advantage because it costs less per GPU, and can therefore provide better performance for the same price. It's nVIDIA's problem for not being price-competitive, not AMD's for having good prices. You have not provided any reasons for why this is unfair, and have not substantiated your opinion with any factual reasons. Your point does not stand.

EDIT: I need to add something: in this day and age, the number of GPUs doesn't mean anything anymore. CrossFire and SLI scale so well in mainstream titles that you literally have 3x a single GPU (almost, not quite). The number of chipsets is inconsequential, it just depends on how much they cost.
 
I don't want [H] to cover anything additionally, they did price comparison and I agree with their comparison. I'm not bashing the writers, the site or anyone. All I'm saying is that I like to see Tri vs Tri outside price point comparison. Nobody is obligated to do it and I have no problem with it. However, it should also be fine for me to agree with this review and site something that I also like to see too (without demanding it be done). I haven't done anything out of the ordinary, I see way too much fuss over a small matter.

fyi im agreeing with you, im not trying to make it a big deal.
 
Out of curiousity, do you go on car forums and complain when the performance of two cars are compared against eachother that are in the same class: Ie Luxury and same price range: Ie both cost $26,000.00 only because one company offers a V6 at that price and the other company offers a V4? Do you complain you'd like them to load a custom bios onto the fuel injectors to disable 2 pistols onto the V6 so the comparison of performance is more 'fair' regardless of the fact that if a customer spent the same $26,000.00 the actual car they'd get would have 6 cyclinders? Wouldn't that just promote the adding of mis-information?

Sorry, but, most people who quad-sli or tri-sli have money. People with money buy top end-cards. They don't buy 2nd tier cards x 3. They'd rather buy top tier x2 b/c top tier cards have more ram on their GPU generally providing more performance at high resolution/high levels of AA which is the purpose of quad/tri-SLIing to begin with. The number's your requesting for tri-core versus tri-core comparison of 3x570 versus 3x6970s would be a waste of the reviewers time as those who could afford 3x570 would realize 2x580 is better performance and buy that instead.

I'm not talking about price perspective. It's always fine to do that however, but if I was looking at a review of V8 engines cars for instance, I'd like to at least see something comparing all V8's without having to stick in a V10 or V6 car into the mix. They are free to do it in price comparisons though, no problem there at all. It's always good to see both or multiple views. However, you can't arbitrarily stick different cars in and insist people only choose a car by price. There are many factors that go into decisions like that which can't just be disregarded.

The 6900 setup doesn't have a clear advantage because it has more GPUs, it has a clear advantage because it costs less per GPU, and can therefore provide better performance for the same price. It's nVIDIA's problem for not being price-competitive, not AMD's for having good prices. You have not provided any reasons for why this is unfair, and have not substantiated your opinion with any factual reasons. Your point does not stand.

EDIT: I need to add something: in this day and age, the number of GPUs doesn't mean anything anymore. CrossFire and SLI scale so well in mainstream titles that you literally have 3x a single GPU (almost, not quite). The number of chipsets is inconsequential, it just depends on how much they cost.


You guys are still completely ignoring everything besides price point. I'm not talking about that, I already said I agree with it from a price perspective, I just don't from any other perspective. I don't see what the big deal is with having price comparison and looking at other comparisons that show Dual vs Dual, Tri vs Tri as well etc. Once again, why stamp out stuff you don't want to see, nothing is wrong with looking at both. If you completely want to suppress other views and insist everyone go by one standard, then that's where the most apparent form of "unfairness" comes in.
 
Regarding the car comparison, you could measure many things about a car - top speed, horsepower, torque, gas mileage, acceleration, etc. The number of cylinders does not matter, all other things being equal. Think about it: if two cars have the exact same specs but different number of cylinders, does is really make a difference? If one car beats the other in every category but it has more or less cylinders (pick your preference) does it really make a difference? Sure, if a car has less cylinders but the same specs then you get bragging rights, but in the end it doesn't matter.

Just food for thought.
 
You guys are still completely ignoring everything besides price point. I'm not talking about that, I already said I agree with it from a price perspective, I just don't from any other perspective. I don't see what the big deal is with having price comparison and looking at other comparisons that show Dual vs Dual, Tri vs Tri as well etc. Once again, why stamp out stuff you don't want to see, nothing is wrong with looking at both. If you completely want to suppress other views and insist everyone go by one standard, then that's where the most apparent form of "unfairness" comes in.

What other perspective would be more important to the end user besides price and performance? Heat, slot requirements and noise might affect your purchasing decisions, points which have been addressed by the article in passing. And yet you continually push the dual-gpu versus single-gpu perspective, even though the number of gpus on a SINGLE card would be transparent to the end-user in this case.
 
What other perspective would be more important to the end user besides price and performance? Heat, slot requirements and noise might affect your purchasing decisions, points which have been addressed by the article in passing. And yet you continually push the dual-gpu versus single-gpu perspective, even though the number of gpus on a SINGLE card would be transparent to the end-user in this case.

Considering he is a noobie and signed up 3 days ago. You have to think he is one of those people paid by Nvidia to cause problem in hardware forums.

It is a known practice from Nvidia.

If he wants tri vs tri. He needs to find another article somewhere else that has what he wants.

This article was $1000 of Nvidia cards vs $1000 of AMD cards.

Not AMD's fault they have a better price/performance Ratio.
 
Considering he is a noobie and signed up 3 days ago. You have to think he is one of those people paid by Nvidia to cause problem in hardware forums.

It is a known practice from Nvidia.

You are right. Didn't taught about that. I'm also a part time audio/video reviewer for a french AV magazine, and we had a really bad cases of infiltration from people working for a high-end pre/pro company, and trying to put down other pre/pros in the same price range (7K to 10K$). Alot of money. The owner of the magazine/forum, a friend of mine, caught them by chance. We would never have knowned...It almost went in court.

You never think that big company would do this, but they do. Sad but true. I was even offered freebies and money to make a ''positive'' review for a 45K$ projector. :confused:

Like I always say now since the pre/pro incident, you never know who's hiding behind the keyboard. Sadly.

Not AMD's fault they have a better price/performance Ratio.

No. But Nvidia's fault for overcharging and asking 500$ for the 580 with only 1.5Gb of VRAM.
 
he signed up here to argue about this until his chest stops hurting, the same as he did over at OCN. just totally derailed the thread. everyone ignore him and hope for a ban. he is not here to "disucuss" anything.
 
the number of gpus on a SINGLE card would be transparent to the end-user in this case.

Thanks, this is what I was trying to explain to the person ^^^, you wrote it very concisely. It really doesn't matter whether you have three of two chipsets. It doesn't matter whether you have a V6 Turbo or V8: if the user's experience isn't noticeably impacted. Of course, the car example isn't good because turbo lag is noticeable, but you get the point. It's not the means that counts, it's the end result.

EDIT: And we're comparing multiGPU setups, it's not like one has a big advantage for not having to use multiGPU. The person ^^^ has only stated a view out of personal opinion, there is no factual base supporting his ideas. I don't care to stamp out anyone's ideas, I tell people when they don't make sense or are beside the point.
 
Considering he is a noobie and signed up 3 days ago. You have to think he is one of those people paid by Nvidia to cause problem in hardware forums.

It is a known practice from Nvidia.


*facepalm*

Was wondering why nothing was making sense in this thread... should have remembered...

bad product = covert marketing

part of basic marketing 101...


sigh...
 
Knowing the performance and price difference I don't understand how anyone could buy a Nvidia card. Unless you're running 3D vision why would you buy a Nvidia card?
 
Did I just read that right? CPU limited with an overclocked i7 @ Eyefinity resolutions, scary :eek:. Looks like these cards will need the push of a Bulldozer to get em moving. :D
As always, great review. Thanks.

Im super late on this.....but holy wow.
 
Ok i have a new setup i72600k @4.8ghz and trifire 6950/70 running at 8x8x4x pcie 5760x1200res

Ok the game runs ok with 3 cards with good fps but as soon as i up the AA from 1x the game lags in sensertivity from the mouse to screen but the fps are still good, ok so if i disable the 3rd card and run crossfire the game will play super smooth with 8 x AA and still good fps, is this a Battlefield game problem, Drivers or the fact my 3rd card is running at 4x pcie, the 3 cards run fine in benchmarks! not had chance to try with a different game yet. Thanks for any help.. ( noticed a sim problem they had in this review)
 
I was reading around other forums, seems that Corsair HX850 can run this setup, is that true? lol.
 
Back
Top