***Official GeForce GTX660/GTX650 Review Thread***

[Elfear]

How can you compare a 8970 its not even out yet? The 7970 Ghz would be the last generation of the 7 series, not really a midrange. Who knows if they will make a 8990 after that or a 8970Ghz lol.

Exactly. Very much like the mythical GK110 gaming card. That's the analogy thilanliyan was trying to make.

 

[BallaTheFeared]

Exactly. Very much like the mythical GK110 gaming card. That's the analogy thilanliyan was trying to make.

There is nothing mystical about it, weather it comes to desktops or not doesn't change what GK104 is. Not to mention adding a 8990 dual gpu card and the "GHz" card into it shows both him and you seemingly are unaware of what is actually being discussed.

Did placing two GF104 cores on a 2Win card morph it into a GF110 core? No, that's stupid, similar to what was quoted. It's simply two GF104 cores, mid range as it were, that produced similar performance to the GTX 580.

A 7990 isn't a new gpu, it's two Tahiti gpus on one card. I can't believe I have to explain this.

Fermi GF104 vs GF100

Kepler GK104 vs GK110

What's so hard about this?

 

[cusideabelincoln]

Wth? I am not labeling RS as pro or anti anything. Why don't you read the whole thread and come to your own conclusions about what really happened: http://forums.anandtech.com/showthread.php?t=2247849

Basically RS was talking smack about GCN arch and saying Kepler was more futureproof, and now it appears he's done a 180 even going so far as to use the same Unigine argument I made four months ago but which he shrugged off four months ago.

You talk about DC but look at his comments about MSAA and tess and FP16.

My point is not that RS was/is wrong, and sorry to RS if you think I'm picking on those posts of yours yet again. My point is that things change--sometimes quite quickly in favor of one arch or the other--so don't get too hung up on trying to buy something you think will be more futureproof than the other company's card. They will both be obsolete soon anyway. Can't really futureproof in a fast-moving industry like GPUs.

You clearly missed the point I was making, which is basically the same one you just made. RS was wrong back then - so yeah we agree on this. I saw his comments on tess and FP16. So what? I already addressed that. He was using erroneous and/or incomplete data, but by his verbose nature he tries to extrapolate too technically. One thing then leads to another, and he's going down the wrong path.

He was wrong then, but he's not here today beating the same drum. He admits he was wrong. Is he wrong today still, for different reasons? That is up for debate, and since he uses so much data from different sources you basically do have to pick apart every argument as you need to analyze each piece of raw data he uses to form his conclusions. In order for him and anyone else to come up with a theory, people need to understand just exactly what the data is and is not showing you. You have to look deeper than the surface. And the prime example of that is RS using a benchmark of the DEMO of Dirt Showdown to come to his conclusion about FP16 textures. I think it's obvious now that the Showdown Demo wasn't using DC. Since the favor swung so far in AMD's favor once the retail game shipped, one can now theorize that DC is a bigger competitive advantage for AMD than FP16 is for Nvidia.

Also I'll clarify that "labeling" him pro-anything wasn't accusatory towards you. Many other users want to label him pro-whatever, and I took your outright use of the term "pro-AMD" as an opportunity to address why no one should be doing that.

 

[sze5003]

There is nothing mystical about it, weather it comes to desktops or not doesn't change what GK104 is. Not to mention adding a 8990 dual gpu card and the "GHz" card into it shows both him and you seemingly are unaware of what is actually being discussed.

Did placing two GF104 cores on a 2Win card morph it into a GF110 core? No, that's stupid, similar to what was quoted. It's simply two GF104 cores, mid range as it were, that produced similar performance to the GTX 580.

A 7990 isn't a new gpu, it's two Tahiti gpus on one card. I can't believe I have to explain this.

Fermi GF104 vs GF100

Kepler GK104 vs GK110

What's so hard about this?

I threw the 8990 and Ghz in there for shits and giggles not because I don't know what's going on I just don't see the comparison as important, my opinion.

 

[BallaTheFeared]

Yeah I'm lost as to what the problem is and what comparison you're speaking to. The whole 8970 thing was just redic.

 

[blastingcap]

You clearly missed the point I was making, which is basically the same one you just made. RS was wrong back then - so yeah we agree on this. I saw his comments on tess and FP16. So what? I already addressed that. He was using erroneous and/or incomplete data, but by his verbose nature he tries to extrapolate too technically. One thing then leads to another, and he's going down the wrong path.

He was wrong then, but he's not here today beating the same drum. He admits he was wrong. Is he wrong today still, for different reasons? That is up for debate, and since he uses so much data from different sources you basically do have to pick apart every argument as you need to analyze each piece of raw data he uses to form his conclusions. In order for him and anyone else to come up with a theory, people need to understand just exactly what the data is and is not showing you. You have to look deeper than the surface. And the prime example of that is RS using a benchmark of the DEMO of Dirt Showdown to come to his conclusion about FP16 textures. I think it's obvious now that the Showdown Demo wasn't using DC. Since the favor swung so far in AMD's favor once the retail game shipped, one can now theorize that DC is a bigger competitive advantage for AMD than FP16 is for Nvidia.

Also I'll clarify that "labeling" him pro-anything wasn't accusatory towards you. Many other users want to label him pro-whatever, and I took your outright use of the term "pro-AMD" as an opportunity to address why no one should be doing that.

I read what you wrote as your saying I was labeling him (RS) as pro-AMD, when I did not. I said his post was pro-AMD.

Again, a reminder: this thread arc started with silverforce claiming NV's only advantage was Physx which I'm sure most people would not agree with, and from there RS wrote counterpoints, but that's beside the point--I was simply disproving silverforce's post, NOT arguing that AMD or NV was better.

In later posts I also made the argument that which arch has more of an advantage can shift quickly and that it's not a good idea to try to futureproof based on minor and possibly temporary advantages. Heck, even if arch stayed constant, node sizes keep dropping, so it's STILL not a good idea to futureproof if it's too expensive... so long as node size drops translate to higher perf/price ratios of course, which they might not do as much in the future due to the escalating expense of building more advanced fabs.

Enough with this thread arc, which only started with Silverforce's claim that NV's only advantage was Physx, something I think most people would disagree with. Back to the OP's thread topic, please.

 

[thilanliyan]

No that would make it a last gen card, the fact that this needs to be said pretty much sums up Anand.

So if nvidia calls gk110 a GTX780 will you still say it is the high end of the GTX6xx generation? According to your logic, a higher sequence number implies a different generation correct?

Makes no sense to claim a phantom card is the high end..is there even any confirmation from nvidia that gk110 is being released for desktops? If there isn't then my statement is as valid as yours.

 

[BallaTheFeared]

So if nvidia calls gk110 a GTX780 will you still say it is the high end of the GTX6xx generation?

Makes no sense to claim a phantom card is the high end..is there even any confirmation from nvidia that gk110 is being released for desktops? If there isn't then my statement is as valid as yours.

I will say it's the high end of Kepler, much like GF100 and GF110 are the high end of Fermi.

I will say Nvidia failed to deliver a high end 6xx series product.

 

[Silverforce11]

I will say it's the high end of Kepler, much like GF100 and GF110 are the high end of Fermi.

I will say Nvidia failed to deliver a high end 6xx series product.

That's just your opinion.

They delivered high-end where its manufacturable. GK104 is the best thing they can make this year on 28nm, its their top-dog regardless of your or my opinion that its a mid-range GPU being sold at rediculous prices.

 

[boxleitnerb]

HD7970 GE in that test is a reference card. No such card exists in retail channels.

IF you have been reading this forum and paying attention, HD7970 @ 1150-1165mhz draws about 225-238W at full load, while after-market 7970 GE cards draw about 200W or less.

You guys keep linking HD7970 GE in reviews and we repeated for 20-30x that it's meaningless and you still keep linking it. :hmm:

Where can I buy a reference 1.25V HD7970 GE card? Please let me know.

So specially binned chips for one partner model (which is not the GHz Edition btw as it was released waaaay before that), possibly with lower voltage, will result in lower power consumption? What a surprise. Please show me at least 5 GE partner cards that have such a low consumption. I have searched the web and found several reviews of the GHz Editions from the partners. ALL of them consume basically what the reference design consumes:

Club3D RoyalAce Edition, uses more:
http://www.fudzilla.com/home/item/2...d-7970-ghz-royalace-edition-reviewed?start=10

Sapphire Toxic GHz Edition, uses more:
http://www.hardwareluxx.de/index.ph...toxic-radeon-hd-7970-ghz-edition.html?start=5

His 7970 X Turbo Edition/GHz Edition, uses more:
http://www.guru3d.com/article/his-radeon-hd-7970-x-turbo-edition-review/8

Gigabyte 7970 GHz Edition (unfortunately I could not find a review, but Gigabyte recommends a beefier power supply. That would certainly not be the case if the card used the same or less):
http://videocardz.com/33890/gigabyte-launches-radeon-hd-7970-ghz-edition-with-windforce-3x-cooling

You can find GTX680 OC models that use less or the same as the reference model, too:

uses the same:
http://www.techpowerup.com/reviews/MSI/GTX_680_Twin_Frozr_III/26.html

uses less:
http://www.techpowerup.com/reviews/Palit/GeForce_GTX_680_Jet_Stream/1.html

So your example with the VTX3D (which is not a GHz Edition), but a custom OC Edition, is meaningless.

If AMD are stupid enough to send an inefficient card to reviewers that will (mostly) never see the light of day in that form, it is their fault. But I've come to expect nothing else from AMD marketing. And at least in case of Club3D, there exists a card in reference design:
http://legitreviews.com/news/13495/

Probably got dumped later on and replaced by the RoyalAce Edition.

 

[Red Hawk]

As I said in another thread. Nvidia supposedly failing to release GK110 as their high-end part is their problem. And really, it seems at some point Nvidia decided that GK104 would be released as their high-end part instead. It was released as the GTX 680, the obvious successor to the GTX 280/480/580 line of enthusiast graphics cards. And they aren't doing what AMD did when they released the 6800/6900 series because they already released the GTX 690 as the dual GPU product.

The reality is that, whatever it was originally conceived as, GK104 is Nvidia's enthusiast part for the Geforce 600 generation. GK106 is the mainstream chip while GK107 is the budget chip. No point in talking as if that's not the case.

 

[AtenRa]

Why don't we compare the 7970 against the GTX680, which cost nearly the same in March?

You got to be kidding right ??
Why not compare the HD7970 to GTX580 in January which they cost nearly the same ???

 

[Grooveriding]

So specially binned chips for one partner model (which is not the GHz Edition btw as it was released waaaay before that), possibly with lower voltage, will result in lower power consumption? What a surprise. Please show me at least 5 GE partner cards that have such a low consumption. I have searched the web and found several reviews of the GHz Editions from the partners. ALL of them consume basically what the reference design consumes:

Club3D RoyalAce Edition, uses more:
http://www.fudzilla.com/home/item/2...d-7970-ghz-royalace-edition-reviewed?start=10

Sapphire Toxic GHz Edition, uses more:
http://www.hardwareluxx.de/index.ph...toxic-radeon-hd-7970-ghz-edition.html?start=5

His 7970 X Turbo Edition/GHz Edition, uses more:
http://www.guru3d.com/article/his-radeon-hd-7970-x-turbo-edition-review/8

Gigabyte 7970 GHz Edition (unfortunately I could not find a review, but Gigabyte recommends a beefier power supply. That would certainly not be the case if the card used the same or less):
http://videocardz.com/33890/gigabyte-launches-radeon-hd-7970-ghz-edition-with-windforce-3x-cooling

You can find GTX680 OC models that use less or the same as the reference model, too:

uses the same:
http://www.techpowerup.com/reviews/MSI/GTX_680_Twin_Frozr_III/26.html

uses less:
http://www.techpowerup.com/reviews/Palit/GeForce_GTX_680_Jet_Stream/1.html

So your example with the VTX3D (which is not a GHz Edition), but a custom OC Edition, is meaningless.

If AMD are stupid enough to send an inefficient card to reviewers that will (mostly) never see the light of day in that form, it is their fault. But I've come to expect nothing else from AMD marketing. And at least in case of Club3D, there exists a card in reference design:
http://legitreviews.com/news/13495/

Probably got dumped later on and replaced by the RoyalAce Edition.

Who cares about all that though ? Hasn't power consumption always been the price for putting out the fastest single GPU card on the market ?

AMD upped power consumption and regained the performance crown. You've been able to buy 1GHz, 1.1 GHZ and 1.2GHZ 7970s for a while now. Power consumption has always been the price of having the fastest card available.

Don't you still run 580 Tri ? I would of figured you knew that and could care less about power consumption with 3 power hungry hot cards like that.

Power consumption is irrelevant to enthusiasts buying the fastest cards available. Many don't even care about noise levels, some do. There are plenty of aftermarket cards or the choice to water cool to run power hungry cards without dealing with noise levels.

http://www.newegg.ca/Product/Product...82E16814125439

http://www.newegg.ca/Product/Product...82E16814131471

http://www.newegg.ca/Product/Product...82E16814127670

http://www.newegg.ca/Product/Product...82E16814129265

Plenty.

 

[boxleitnerb]

You have to differentiate here.

We were discussing architectures, not personal preferences or tolerances when it comes to power. I was correctly saying that you have to observe technical parameters like what a card does with the bandwidth, compute power and power consumption it has. That is (amongst other things) what defines an architecture.
I admit that my initial musings about the 7970 GE power consumption were not really relevant because this card only highlights that you can make any card/architecture inefficient if it leaves its optimal perf/W window when using high voltage due to yield reasons; that has nothing to do with architecture. In context of that discussion, the 7970 GE is irrelevant.

The 680 was pushed to compete with a card (regular 7970) that has 30% higher bandwidth and theoretical compute performance. Keeping that in mind, I believe GK104 fares quite well. It loses in some cases because it has less raw power, but it keeps up or "wins" (slightly, that is) in the majority of other cases.

 

[BallaTheFeared]

That's just your opinion.

They delivered high-end where its manufacturable. GK104 is the best thing they can make this year on 28nm, its their top-dog regardless of your or my opinion that its a mid-range GPU being sold at rediculous prices.

Bwhahaha, ok, whatever you say. I can understand your need to defend a purchase.

GK104 represents several things, none of which you'll concede to.

 

[antihelten]

Club3D RoyalAce Edition, uses more:
http://www.fudzilla.com/home/item/2...d-7970-ghz-royalace-edition-reviewed?start=10

Sapphire Toxic GHz Edition, uses more:
http://www.hardwareluxx.de/index.ph...toxic-radeon-hd-7970-ghz-edition.html?start=5

His 7970 X Turbo Edition/GHz Edition, uses more:
http://www.guru3d.com/article/his-radeon-hd-7970-x-turbo-edition-review/8

Gigabyte 7970 GHz Edition (unfortunately I could not find a review, but Gigabyte recommends a beefier power supply. That would certainly not be the case if the card used the same or less):
http://videocardz.com/33890/gigabyte-launches-radeon-hd-7970-ghz-edition-with-windforce-3x-cooling

Just so you know, non of the cards you listed actually runs at the standard 7970 GHz boost clocks (ie. 1050MHz). The club-3d card boosts to 1100, the Sapphire to 1200, the HIS to 1180, and the Gigabyte is clocked at 1100 (it doesn't appear to actually boost, according to it's specs)

 

[boxleitnerb]

The Sapphire Toxic boosts to 1050 at default. Only with the second bios it boosts higher.
Anyway, the post was just to clarify that the partner-made GHz Editions use quite a bit more than the ref design and not less as RS wanted to make us believe.

 

[Ferzerp]

You clearly missed the point I was making, which is basically the same one you just made. RS was wrong back then - so yeah we agree on this. I saw his comments on tess and FP16. So what? I already addressed that. He was using erroneous and/or incomplete data, but by his verbose nature he tries to extrapolate too technically. One thing then leads to another, and he's going down the wrong path.

He was wrong then, but he's not here today beating the same drum. He admits he was wrong. Is he wrong today still, for different reasons? That is up for debate, and since he uses so much data from different sources you basically do have to pick apart every argument as you need to analyze each piece of raw data he uses to form his conclusions. In order for him and anyone else to come up with a theory, people need to understand just exactly what the data is and is not showing you. You have to look deeper than the surface. And the prime example of that is RS using a benchmark of the DEMO of Dirt Showdown to come to his conclusion about FP16 textures. I think it's obvious now that the Showdown Demo wasn't using DC. Since the favor swung so far in AMD's favor once the retail game shipped, one can now theorize that DC is a bigger competitive advantage for AMD than FP16 is for Nvidia.

Also I'll clarify that "labeling" him pro-anything wasn't accusatory towards you. Many other users want to label him pro-whatever, and I took your outright use of the term "pro-AMD" as an opportunity to address why no one should be doing that.

A lot of you are falling for the old "firehose of information must be correct" trick. Making massive posts that link to only tangentially relevant information does not make a person correct. At times, it's just one big smoke screen to make a person seem to be more well reasoned than they are. Half the time, the guys posts go off on such tangents, but people assume lots of words and pictures means a person is correct.

When your position is correct and well founded, often times you need far, far fewer words/pictures/time to present it than someone who is trying really, really hard to make everyone else agree with them regardless of merit. With this guy, it's not pro this or pro that. It is more, I have an opinion, and through sheer verbosity, I will try to force you to agree with it. However, unlike impartial reviewers, etc. He tries to present only information he feels agrees with his view of the month, and since he's regurgitating pictures from every site in the world, there are plenty of pictures to support his current position, and plenty to leave out because they don't. Interestingly, when AT, HOCP, etc are all saying one thing, and he's saying the opposite, why would I trust his word? Because lots of pictures? Really?

 

[antihelten]

The Sapphire Toxic boosts to 1050 at default. Only with the second bios it boosts higher.
Anyway, the post was just to clarify that the partner-made GHz Editions use quite a bit more than the ref design and not less as RS wanted to make us believe.

They tested the card using the second bios. As such your post only shows that partner-made GHz edition cards clocked/boosting above and beyond the normal GHz clocks, use more power than the ref design.

I believe that the point that RS was trying to make was that partner-made cards use less power than the reference version given similar clocks (mainly due to lower voltage).

 

[railven]

A lot of you are falling for the old "firehose of information must be correct" trick. Making massive posts that link to only tangentially relevant information does not make a person correct. At times, it's just one big smoke screen to make a person seem to be more well reasoned than they are. Half the time, the guys posts go off on such tangents, but people assume lots of words and pictures means a person is correct.

When your position is correct and well founded, often times you need far, far fewer words/pictures/time to present it than someone who is trying really, really hard to make everyone else agree with them regardless of merit. With this guy, it's not pro this or pro that. It is more, I have an opinion, and through sheer verbosity, I will try to force you to agree with it. However, unlike impartial reviewers, etc. He tries to present only information he feels agrees with his view of the month.

Well that is a very negative generalization. Some people are overly interested in the subject matter that they over-complicate explanations with too much information. Working with doctors, I run into that situation far too often. A simple question turns into an hour lecture as they try to explain something to me in the finest of details.

Now, I'm not saying that is what RS is doing here, but by your explanation here - my doctors are wrong. I'll let them know that.

What I know about reading these forums (and often it seems I'm probably the only one really reading them haha) is - people respond to multiple points in one post and often the original statement they are responding to gets lost. I think this causes a lot of the tension, as Poster A is responding to Poster B, C, D and E but only quoted Poster F and then you'll see a follow-up by Poster F saying he never said any of the things Poster A is responding to.

It's just amusing reading people's opinions about RS. When he says glowing things for nVidia certain people (and a forum) say nothing. When he's saying glowing things about AMD certain people (and a forum) say a lot.

This place is too invested in people's opinions around here. It's like - they just HAVE to defend the products. I get why Keys has to, but some of the other posters (from both sides) - really? Sometimes its worse than sport fans haha.

 

[Ferzerp]

I pointed out that in the last thread, it wasn't the nvidia folk that crapped it up, it was the amd folk.

His response? Quoting me and some tangential and totally unrelated rant about AIBs and how it is all nvidia's fault.

If you look closely, if he can't support his position of the month, he tries to shift the argument in to another area and throw out so many barely related links that one cannot really discuss anything with the guy due to the firehose effect.

In fact, those doctors are potentially just trying to get you to shut up and do/believe what they say (whether they are right or wrong). I do the same thing at work. When someone pushes, and I don't have time to deal with it, it is wall of information time to quell all argument. At the end of it, they feel no differently, but they've been intimidated and will go away and leave me alone.

You'll also notice the attempts to dictate other's behavior "no don't quote all that" etc. It's almost like it is an attempt to force the forum in to his particular model of what it should be and any dissent is unwelcome. (Ever try to edit a wikipedia page that has someone very invested in it?)

 

[blackened23]

Summary of this thread:

 

[railven]

I pointed out that in the last thread, it wasn't the nvidia folk that crapped it up, it was the amd folk.

His response? Quoting me and some tangential and totally unrelated rant about AIBs and how it is all nvidia's fault.

If you look closely, if he can't support his position of the month, he tries to shift the argument in to another area and throw out so many barely related links that one cannot really discuss anything with the guy due to the firehose effect.

In fact, those doctors are potentially just trying to get you to shut up and do/believe what they say (whether they are right or wrong). I do the same thing at work. When someone pushes, and I don't have time to deal with it, it is wall of information time to quell all argument. At the end of it, they feel no differently, but they've been intimidated and will go away and leave me alone.

You'll also notice the attempts to dictate other's behavior "no don't quote all that" etc. It's almost like it is an attempt to force the forum in to his particular model of what it should be and any dissent is unwelcome. (Ever try to edit a wikipedia page that has someone very invested in it?)

Gotcha. Guess, I'll just leave it at that.

 

[Termie]

...

Enough with this thread arc, which only started with Silverforce's claim that NV's only advantage was Physx, something I think most people would disagree with. Back to the OP's thread topic, please.

Seriously, guys, this.

I don't think there's been a comment about the 650 or 660 in three pages or so. I'd rather not see this thread just become a 680 vs. 7970 slugfest.

I'll go ahead and throw out one major advantage of the 660. It is nearly as fast as the 7870, but is available in a 7 inch model from Zotac, whereas the shortest 7850 is the 7.7 inch MSI TF3, while the shortest 7870 appears to be around 10 inches.

For someone using a SFF case, that's a huge win for Nvidia. I'd love a 7870, which appears to have similar power consumption compared to the 660, but AMD partners really have to find a way to get these GPUs on smaller PCBs.

 

[railven]

Summary of this thread:

Pretty much all the threads around here. Defending someone's product of choice seems to matter more than the product itself.

Oh well, least it helps pass the time at work