Nvidia BigK GK110 Kepler Speculation Thread

~513mm² die area.

Lol you are fast, i was ask me how many time it will take for someone to come with a die size estimation.

By the way, that's rather surprising to see big high-end GPU from NV without protective cap this time. Looking at the board itself and the bare die, it looks like this GeForce SKU was a quick Tesla model convert for the consumer market. I wonder if some HW or BIOS mods could be done here. ;)

Do you mean for reenable feature disabled ? Huum, looking the card will be faster, and cost way lot under the Tesla K20 ... im sure they have take no risk with that.
 
Last edited by a moderator:
It's beautiful card and clearly follows the design language of the GTX 690 cooler and face it, aluminium and magnesium is better materials than any 3rd party plastic shrouds from AIBs.

Also endless whining about price, yet many people still pay over $1000 for multi GPU cards with micro-stuttering and scaling issues. With this card, you don't have to deal with micro-stuttering and scaling issues and game incompatibilities inherent with multi GPU.

If you can't afford it, stick to the $200-300 price range cards which gives good bang for buck like most people do.
 
It's beautiful card and clearly follows the design language of the GTX 690 cooler and face it, aluminium and magnesium is better materials than any 3rd party plastic shrouds from AIBs.

Indeed, it look absolutely beautifull.. I allways like industrial style. Well i allways use h2o, so its not like it will stay like that a long time if i buy this card.
 
Last edited by a moderator:
Also endless whining about price....

Seriously. Chalnoth touched on this before but this price/performance "fairness" nonsense is imaginary. There are a lot of reasons why people drop dough on expensive shit, including the simple fact that they can afford to.

If a company wants to target that market there's nothing wrong with that. They will adjust prices to get demand to the right level. Just know that they're not targeting you and quit bitching about it.
 
I don't get it. One second he is whining about low end "affordable" junk, the next everything is too expensive. Oh, maybe I do get it...
 
If those performance numbers hold true I don't understand how nVidia could possibly justivy that 900-1000€ pricetag on this thing

Probably because there's no business reason to charge less? Just like there was no justifiable business reason for AMD to charge less than 550 USD for the 7970 when it launched?

You price it based on how much you can supply combined with how much you estimate demand will be. In both cases charging less doesn't mean more people will buy the card. After all 7970 was sold out for months at that price. Titan is likely to be sold out at the rumored availability as well. And I'm sure Nvidia doesn't want a repeat of GTX 680, where the card was sold out for quite a long time. They probably wished they had priced it higher after seeing the level of demand for the card.

Manufacturing capacity for GK110 likely means that it cannot meet high demand, thus the smartest thing to do is price it at a point where you think demand will roughly equal your ability to supply product.

It's not about pricing it so that everyone can afford it. It's all about pricing it for maximum profit with ability to produce enough chips to meet the demand at the targeted price point.

Personally, I think the price is silly for a card that will be used by most for gaming. But I'm willing to bet there's enough people willing to pay that price that Nvidia won't have trouble selling the limited supply that they'll have for the consumer market. I'm guessing that the majority of GK110 cores will be allocated to the professional market.

Regards,
SB
 
What are the technical issues which prevent them from manufacturing 10x the wafers allocated for GK110, and thus reducing for example the price by 30%?
 
The NDA is supposed to expire tomorrow, isn't it? Perhaps we should wait a few hours for the official price before we start complaining about it.
 
So did you post the same statement of "arrogant prices" when AMD had the market to itself for the first 3 months of the HD7970 at $549?

The 7970 was what, ~30% faster than the next fastest card and cost $549. By the looks of it Titan will be ~30% faster than the next fastest card and cost $1000. That's quite a difference.

It's not arrogance though, it simply points to the incredibly low volume of sales they expect for this card.
 
What are the technical issues which prevent them from manufacturing 10x the wafers allocated for GK110, and thus reducing for example the price by 30%?

It's because the market for $700 cards is about as vanishingly small as the market for $1000 cards is, nothing technical about it.
 
What are the technical issues which prevent them from manufacturing 10x the wafers allocated for GK110, and thus reducing for example the price by 30%?

Lower margins. And since Nvidia are a publicly traded company, that's fairly important.

As well if there's a limited number of wafer starts available. Would it be better to dedicate more wafers to GK104, etc.? Or more wafers to GK110 at lower margins than what they would get with a 900+ USD price point?

Also, more wafers to GK110 isn't going to increase your market share significantly, while those wafers going for GK104, etc. will potentially increase your market share quite a bit more.

In other words, would it be a good idea to reduce how many GK104s they make in order to make more GK110s?

Regards,
SB
 
Lower margins. And since Nvidia are a publicly traded company, that's fairly important.
High margins and less $ earned don't trump low margins and more $ earned, especially in publically traded companies. That's how IKEA and Walmart for example grew into the giants they are today (except IKEA isn't publically owned, of course...)

As well if there's a limited number of wafer starts available.
This is a much more likely explanation than a wish to artificially keep margins high at the expense of pure dollar profits.
 
High margins and less $ earned don't trump low margins and more $ earned, especially in publically traded companies. That's how IKEA and Walmart for example grew into the giants they are today (except IKEA isn't publically owned, of course...)


This is a much more likely explanation than a wish to artificially keep margins high at the expense of pure dollar profits.

Walmart and ikea are in very different circumstances than Nvidia. And investors do look at margins, it's a sign of long term profitability, but this really all belongs in a different thread.

Pricing is about what will net them the most money, if they thought selling more Titans at a lower price would net them more money, that's what they would do.
 
Actually, it depends...
If their strategy is focused in other directions and they want to promote tablets and smartphones, for the sake of some weird intentions (here, yes, I assume it's a possibility*), then they will artificially kill people's will to buy their desktop products.
*They do it, by the lack of any innovations in the desktop market, come on, we are still stucked with the stupid 1080p resolution monitors...

I mean- it is essentially price which drives demand for this particular product. I bet that millions of people would have gone for it if it had been, let's say, a 350-400 $ card.

Price is the starting point, not some rough estimations about how many people would go for it.
 
If their strategy is focused in other directions and they want to promote tablets and smartphones, for the sake of some weird intentions (here, yes, I assume it's a possibility*), then they will artificially kill people's will to buy their desktop products.
That doesn't make sense. The whole purpose of spreading products into multiple markets is to protect the company against failure: every company knows that a single product line can go south and lose significant market share (or the market itself could dry up). So it is just good business to diversify into other market segments.

The only reason why a company would pull out of a market they had long done well in would be because they didn't feel that market was worth the investment to product new products for any longer, and that their efforts would have higher payoff elsewhere.

Companies generally will not torpedo the demand of their own products.
 
It's priced to sell at what Nvidia thinks they can get for it at max. There are a variety of factors involved in this. By pricing it at $1000 it gains automatic credibility from those people who fall for the perceived price/quality relationship. That was the market anyway, and chances are they'd pay $1000 or $700. Many buyers of this card will believe it is even better than it is simply because it cost even more than their last one.

It's simple marketing - Nvidia gets an expensive halo product that tops all benchmarks even though it's basically unavailable to 99.9% of gamers. If AMD wasn't so clueless about these things they'd have had an official 7990 in reviewers hands as a counter to it.

edit - it's possible that we'll see a 7990 of some description in tomorrows reviews, but we'll see. :p
 
Last edited by a moderator:
Back
Top