Yes, no sane way to defend that.
Man from Atlantis
Veteran
over such a small chip than 300mm2 i definitely expect more than 50% for Bigboy..
So nVidia ridiculed Tahiti's performance but now they're going to price their supposedly midrange chip at Tahiti levels leaving no room for the "flagship"?
That implies to me that they have nothing faster than GK104 ready and their proper reaction should have been one of relief. Something don't smell quite right and I took a shower today....
That implies to me that they have nothing faster than GK104 ready and their proper reaction should have been one of relief. Something don't smell quite right and I took a shower today....
Nvidia always has room above the current price point. They've done it before, they'll do it again if they feel they can get away with it. The dual cards might limit their ability to charge as much of a premium, but I wouldn't be shocked to see it launch at $699 with dual cards at $749 or whatever.
Nvidia ridiculed Tahiti based on their performance expectations of GK100 because they compared it as enthusiast chip vs enthusiast chip and that is how they assumed they would face off against each other initially.
GK104 was clearly intended to be the performance part, not the enthusiast part. However, AMD left the door open due to Tahiti's mediocre performance.
The $299 claims were probably true and no doubt appeared in official documents until Nvidia knew that Tahiti wasn't meeting their expectations.
AMD has really let Nvidia off the hook here and this allows them to get GK100 ready without rushing it as well. They can take another 6 months, if GK104 is already faster or same performance - and smaller - then it's AMD who has the problem.
GK104 was clearly intended to be the performance part, not the enthusiast part. However, AMD left the door open due to Tahiti's mediocre performance.
The $299 claims were probably true and no doubt appeared in official documents until Nvidia knew that Tahiti wasn't meeting their expectations.
AMD has really let Nvidia off the hook here and this allows them to get GK100 ready without rushing it as well. They can take another 6 months, if GK104 is already faster or same performance - and smaller - then it's AMD who has the problem.
Comedy Gold. There's zero chance that Nvidia was ever going to release a part at $299 (while having several products below that price point launching in a similar time frame) that would replace their entire lineup beyond the $249 price point, but nice try.
and I very much doubt AMD will be sitting on their hands for the next 6 months.
Why not? They almost certainly assumed that AMD would have blown their older chips into oblivion. So why wouldn't they release their smaller midrange chip at $299 when their old chips were (supposed to be) rendered obsolete by AMD anyway? That's the thing here - their older chips are just bad value, they weren't completely blown away.
Nvidia started this generation with the assumption of another price war - anything else would have been a terrible decision.
If they have to spend the next 6 months beating Nvidia's performance card with the enthusiast card they should have started with then it's a pretty hollow victory.
Last edited by a moderator:
No, I'm suggesting that they will price their product where it competes. The only way it would have launched at $299 is if were not much faster than the 560ti. Even if AMD launched no product in the last 4 months, $299 makes no sense for tahiti level performance. You have to realize that if kepler launched at $299 (I'm assuming at least one salvage, slower part at a lower price point) it would be replacing the 560 and everything above it. Meaning their new high end part for the next 6 months would have been $299. It was never going to happen that way.
Assuming it's 6 months until GK100.
You're missing the point anyway - Nvidia did not expect GK104 to be this close to Tahiti, therefore it was always in their minds that it would have been a $300 card. It doesn't matter what it would have been replacing in their own lineup when their expectation would have been that it was going to fall a bit short of Tahiti's performance, and they would also have been anticipating AMD to price the card aggressively.
Before they knew Tahiti's perf, Nvidia probably figured $400-$450 and GK104 +20% performance for the 7970. Why wouldn't they have (based on history), and why wouldn't they have assumed a ~$300 price tag on GK104?
Last edited by a moderator:
You've got way too much guesswork and assumption in there to even bother addressing it. Just ask yourself this, was Nvidia ever going to have their fastest product on the market priced at $299?
Even if GK104 can make money at around $300, the fact is that if performance is neck and neck with 7970, they can sell it at the 7970 price and make a whole lot more money. They'd be financially crazy to charge less from the get go, especially if supply is initially low. It also seems like a good idea to have room to drop prices when GK100/Sea Islands come out.
The 7970 is only 70% faster than the 560Ti at "normal" resolutions. Is it really far fetched to think nVidia was targeting that performance level for the 560Ti replacement?
Still doesn't explain why they would price it that high if there was a bigger chip coming soonish. That tells me there's no big chip arriving soon, hence they got lucky.
Granted it's not much but they'll still have the 590 holding a premium.
When was the last time Nvidia led with their performance part instead of their enthusiast part though? There are a lot of changes this time around, it shouldn't be a surprise to see Nvidia doing a few things differently.
I agree with trinibwoy that GK100 is probably a bit further than 3 months away, but that doesn't change the possibility that AMD has left an opening here for Nvidia to upgrade the positioning of GK104.
I'm sure you've seen the rumours, 660 ti for the past few months, 670 ti last month, then 680 this month. For me this chip has had it's initial targeted market segment upgraded based on 1) Better than expected performance and 2) Worse than expected AMD performance.
They would be crazy to price the product too high to make any significant revenue. According to John Peddie Research, only 3 million video cards that cost more than $300 were sold last year.
http://alienbabeltech.com/abt/viewtopic.php?f=6&t=24401#p59711
If they were surprised about performance and it's currently about as fast as 7970, it's not as if the surprise was going to be bigger than, say, 20%.
Even with 104 20% slower than 7970, it'd still be faster than a 580. With a 7970 at $550, that would still have left a wide range to price it significantly above $300 and still be an interesting enough deal for a bunch of customers.
Even with 104 20% slower than 7970, it'd still be faster than a 580. With a 7970 at $550, that would still have left a wide range to price it significantly above $300 and still be an interesting enough deal for a bunch of customers.
Silent_Buddha
Legend
Something people should consider is that 28 nm production by TSMC stopped for at least 3 weeks. Pricing a product too low is going to lead to it being out of stock almost constantly which is money left on the table that potentially goes to the competitor.
Well it does still leave them room to price a faster card ate 650-700 USD. Nvidia is no stranger to those price point with graphics cards so I don't see any reasons they wouldn't want to attempt those prices again.
Personally that's WAY out of my price range, but there has historically been people willing to pay that. And if it's a large die on 28 nm. Then pricing it high could possibly be the only way to keep product in stock.
But, if I were to take a guess. I'd agree. No higher performance single chip card from Nvidia for at least 5-6 months.
Eh? Why would they price it lower than a slower card (this presumes that GTX 680 is generally faster)?
Or to put it another way. [strike]Why would they launch GTX 680 at a lower price point than they launched GTX 580 (550 USD) or GTX 480 (550 USD)?[/strike] [edit] Blast my failing old age memory. GTX 480 and GTX 580 launched at 499 USD.
550 USD is already going to be 100 USD lower than they launched GTX 280 (650 USD).
550 USD will be 50 USD cheaper than launch price of the 7800 GTX (600 USD). And that was back in 2005. It'll be similar to what the 7900 GTX launched at, and the 7900 GTX used G71 which was an absolutely tiny chip even for the timeframe.
It's also 50 USD cheaper than the 8800 GTX (600 USD).
Other than the 9800 GTX blip (which basically used the same G92 based chip arch as the 8800 GT), you have to go all the way back to the 6800 Ultra (499 USD) back in 2004 to see Nvidia launching an enthusiast class product at less than 550 USD. [edit again] As well at the GTX 480 (late and coming off the price war that set unrealistically low price expectations from the GTX 280/Radeon 4870 price war) and the GTX 580 (still getting customers used to higher prices again).
Basically anyone who didn't expect 550 USD (or more) as a possible launch price for a GTX 680 just doesn't know Nvidia at all.
Granted there's always the possibility that Nvidia has suddenly decided that it really hates to make money and prices the card lower than what it would otherwise sell at.
Same thing I said about 79xx, 78xx, and 77xx pricing. If GTX 680 is priced too high with regards to the supply of chips, then the price will be adjusted downwards to bring demand more inline with supply. But even then, they will still try to make sure that demand is slightly lower than available supply.
Do I like the fact that enthusiast class cards are for the most part out of my comfortable budget range? Not one bit. But that doesn't prevent me from acknowledging that there are plenty of people willing to pay those prices.
If Nvidia were interested in selling it for cheaper than 550 USD, they would have called it the GTX 660 or possibly GTX 670. I think someone mentioned previously in this thread or another, but usually a company doesn't want to erode the price position of any particular market segment if they can help it. Hence GTX x80 is invariably going to be 550 USD or more going forward if Nvidia has anything to say about it.
You'd be better off arguing that they should have called it the GTX 660. But if there's no GK 100/110 due out in the next 6+ months, they don't really have a choice but to call it GTX 680 if it is faster than the competitions top chip.
Regards,
SB
Well it does still leave them room to price a faster card ate 650-700 USD. Nvidia is no stranger to those price point with graphics cards so I don't see any reasons they wouldn't want to attempt those prices again.
Personally that's WAY out of my price range, but there has historically been people willing to pay that. And if it's a large die on 28 nm. Then pricing it high could possibly be the only way to keep product in stock.
But, if I were to take a guess. I'd agree. No higher performance single chip card from Nvidia for at least 5-6 months.
Eh? Why would they price it lower than a slower card (this presumes that GTX 680 is generally faster)?
Or to put it another way. [strike]Why would they launch GTX 680 at a lower price point than they launched GTX 580 (550 USD) or GTX 480 (550 USD)?[/strike] [edit] Blast my failing old age memory. GTX 480 and GTX 580 launched at 499 USD.
550 USD is already going to be 100 USD lower than they launched GTX 280 (650 USD).
550 USD will be 50 USD cheaper than launch price of the 7800 GTX (600 USD). And that was back in 2005. It'll be similar to what the 7900 GTX launched at, and the 7900 GTX used G71 which was an absolutely tiny chip even for the timeframe.
It's also 50 USD cheaper than the 8800 GTX (600 USD).
Other than the 9800 GTX blip (which basically used the same G92 based chip arch as the 8800 GT), you have to go all the way back to the 6800 Ultra (499 USD) back in 2004 to see Nvidia launching an enthusiast class product at less than 550 USD. [edit again] As well at the GTX 480 (late and coming off the price war that set unrealistically low price expectations from the GTX 280/Radeon 4870 price war) and the GTX 580 (still getting customers used to higher prices again).
Basically anyone who didn't expect 550 USD (or more) as a possible launch price for a GTX 680 just doesn't know Nvidia at all.
Granted there's always the possibility that Nvidia has suddenly decided that it really hates to make money and prices the card lower than what it would otherwise sell at.
Same thing I said about 79xx, 78xx, and 77xx pricing. If GTX 680 is priced too high with regards to the supply of chips, then the price will be adjusted downwards to bring demand more inline with supply. But even then, they will still try to make sure that demand is slightly lower than available supply.
Do I like the fact that enthusiast class cards are for the most part out of my comfortable budget range? Not one bit. But that doesn't prevent me from acknowledging that there are plenty of people willing to pay those prices.
If Nvidia were interested in selling it for cheaper than 550 USD, they would have called it the GTX 660 or possibly GTX 670. I think someone mentioned previously in this thread or another, but usually a company doesn't want to erode the price position of any particular market segment if they can help it. Hence GTX x80 is invariably going to be 550 USD or more going forward if Nvidia has anything to say about it.
You'd be better off arguing that they should have called it the GTX 660. But if there's no GK 100/110 due out in the next 6+ months, they don't really have a choice but to call it GTX 680 if it is faster than the competitions top chip.
Regards,
SB
I think the question of where does GK100 fit is being way too overthought. If it's the fastest single GPU, it will sell for a lot and those buyers aren't that price sensitive to 500 or 700. It's not an accident that this thing is coming 6 months after GK104. Nv knew AMD would have nothing to beat GK100 with so they decided to push with GK104 first. So why give AMD a 6 month run lead with their flagship GPU when the majority of GK100's sales will be in HPC/Pro markets?
Which is why it doesn't need to come before GK104. Which is why it doesn't need to be priced at 500 something.
Which is why it doesn't need to come before GK104. Which is why it doesn't need to be priced at 500 something.
I would expect availability to be low initially (28nm production stoppage, NV overestimating 28nm yield). If that's true and performance is inline with the rumors, then I think there are enough NV fanboys for a GK104 to sell out at $550 short term - they can always drop prices later. (It goes without saying that I hope I'm wrong and the price is below 7970!)
