"The GTX 680 was supposed to be the GTX 660"

[Joseph F]

People, please stop saying this. This is completely unfounded, and almost seems like trolling, at times.
nVidia simply switched to a small-die tactic, like AMD's. If you have evidence that proves otherwise, please speak up.
(And by "evidence", I don't mean "But, but, but, teh die size!!!")

 

[SolMiester]

People, please stop saying this. This is completely unfounded, and almost seems like trolling, at times.
nVidia simply switched to a small-die tactic, like AMD's. If you have evidence that proves otherwise, please speak up.
(And by "evidence", I don't mean "But, but, but, teh die size!!!")

You have evidence of this tactic?....as GK110 is not a small die!

 

[ShintaiDK]

This thread can only go wrong.

 

[tviceman]

People, please stop saying this. This is completely unfounded, and almost seems like trolling, at times.
nVidia simply switched to a small-die tactic, like AMD's. If you have evidence that proves otherwise, please speak up.
(And by "evidence", I don't mean "But, but, but, teh die size!!!")

Nvidia did not switch to a small die strategy. GK110 has been announced. It is kepler's flagship die, in both size and performance. It is around the same size as GF110. GK104 is kepler's second biggest die. GF110 was fermi's flagship die. GF114 was fermi's second biggest die...

If it looks like a duck, and sounds like a duck...

 

[toyota]

what an idiotic thread to start. OP, you cant possibly be that clueless on GK104.

 

[Dark Shroud]

Nvidia did not switch to a small die strategy. GK110 has been announced. It is kepler's flagship die, in both size and performance. It is around the same size as GF110. GK104 is kepler's second biggest die. GF110 was fermi's flagship die. GF114 was fermi's second biggest die...

If it looks like a duck, and sounds like a duck...

Except its only been announced for Tesla cards.

 

[SolMiester]

Except its only been announced for Tesla cards so far

FTFY

 

[Atreidin]

It is a stupid argument anyway. Even if it was true, what would it change? Would Nvidia suddenly have faster stuff? Is it an excuse for Nvidia not being as dominant as they want?

It looks to me like Nvidia just decided to cut off transistors for compute since people buying gaming cards don't care as much about that feature. It gives them lower costs, higher profits.

Then their cultists decided to use that "missing" space as an excuse for why the card isn't "high-end", giving them solace in some coming messiah-card, delivering them to some promised land of douche-baggery.

 

[toyota]

Except its only been announced for Tesla cards.

Nvidia's presentation said Q4 2012 for Tesla K20, and then when asked about a Geforce verison later Huang stated "Q1 2013"

 

[chimaxi83]

There is some evidence to show that GK104 really was meant to be mid range, but the only people that bring this up anymore are trolling NV fanbois who like spending their time here browsing while banned (BWB, feel free to use it ).

My mid range GK104 is performing quite well though, thank you.

 

[nitromullet]

They've announced GK110 for Tesla and launched GK104 as the flagship GTX 680. Given NV's naming convention it seems plausible that GK100 had been planned, but was scrapped for whatever reason. GK104 was probably originally supposed to be their GTX 660, but it also makes a fine GTX 680 that is very competitive with the 7970 so the point is moot IMO.

 

[toyota]

anyone with a clue knows gk104 was not originally intended to be their flagship card. the irony of the op calling someone else a troll while starting a thread like this is hilarious.

 

[Silverforce11]

If it looks like a duck, and sounds like a duck...

It's a Ornithorhynchus anatinus.

 

[BallaTheFeared]

This thread can only go wrong.

It can't possible go wrong, since it started wrong it is just wrong, it can continue to be wrong but it can't go where it started.

 

[DaveSimmons]

It can't possible go wrong, since it started wrong it is just wrong, it can continue to be wrong but it can't go where it started.

No matter where you go, there you are.



I bought my GTX 680 based on its price, performance, power use and noise. Whether it was originally going to be a 660, 670, or 666 doesn't matter to me and I don't see much point in starting a new thread asking us to leave Britney alone.

 

[cmdrdredd]

Yeah it doesn't matter. If it was a gtx 650 and performed like it does I would still buy it. Call it a mid range card if you want. My mid range 670 performs similar to a high end 7970

 

[Homeles]

People, please stop saying this. This is completely unfounded, and almost seems like trolling, at times.
nVidia simply switched to a small-die tactic, like AMD's. If you have evidence that proves otherwise, please speak up.
(And by "evidence", I don't mean "But, but, but, teh die size!!!")

What can be asserted without evidence can be dismissed without evidence. So, there. OP's argument is defeated.

I'm not stupid enough to pay $500 for a 300mm² die. Are you?

 

[Lepton87]

Do you really think that in the planning stages nvidia would allow the possibility of AMD beating their flagship card and decide that increasing memory bandwidth is pointless? Not to mention abandoning compute performance altogether.

 

[SirPauly]

It is what it is! Road maps and planning may indeed change and really irrelevant because the GTX 680 is around 499 MSRP.

My constructive nit-pick was price/performance based on this is a substantial and significant node and arch change -- these are indeed rare. Based on comparisons with history this was not substantial and significant price/performance and felt more incremental and evolutionary.

 

[Atreidin]

What can be asserted without evidence can be dismissed without evidence. So, there. OP's argument is defeated.

I'm not stupid enough to pay $500 for a 300mm² die. Are you?

I wasn't aware that these things sold in units of area. It should be so much simpler now to design everything! They can fire so many engineers, thus lowering cost even more! Genius!

I'd like 10 square centimeters of GPU, hold the mayo.

 

[HurleyBird]

double post

 

[HurleyBird]

It's impossible to know. Really depends whether or not there was supposed to be a GK100. In any case, this thread is the perfect fanboy detector

 

[ShadowOfMyself]

The point is that this is irrelevant... Its the fastest card Nvidia has, regardless of what it is called, so its their high end

Because by the die size/power logic, every previous AMD card since the 3870 has been a midrange card

Funny to see NV fanboys spinning stuff all the time though... By the time GK110 is here, so will the 8970, so whats your point?

 

[NIGELG]

People, please stop saying this. This is completely unfounded, and almost seems like trolling, at times.
nVidia simply switched to a small-die tactic, like AMD's. If you have evidence that proves otherwise, please speak up.
(And by "evidence", I don't mean "But, but, but, teh die size!!!")

But you just said it.

 

[Homeles]

I wasn't aware that these things sold in units of area. It should be so much simpler now to design everything! They can fire so many engineers, thus lowering cost even more! Genius!

I'd like 10 square centimeters of GPU, hold the mayo.

A wafer is only so big, genius.

The point is that this is irrelevant... Its the fastest card Nvidia has, regardless of what it is called, so its their high end

Because by the die size/power logic, every previous AMD card since the 3870 has been a midrange card

Funny to see NV fanboys spinning stuff all the time though... By the time GK110 is here, so will the 8970, so whats your point?

"Mid range card" is completely dependent on what cards are available in the market segments above and below it.

The top parts from HD 3000 through HD 7000 have all been mid sized dies, as far as GPUs go.