NVIDIA GF100 & Friends speculation

AlphaWolf said:
What evidence to the contrary? There's a whole lot of supposition, no evidence.
No evidence???

How about this: consistently high gross margins and a respectable 8% profit.

There are no hard numbers at all that support the other sides' argument.
 
Last edited by a moderator:
No. The original claim is that someone said GTX 460 was/is selling at loss. Then someone came in claiming that NVIDIA is selling "everything" with profit. Both at not necessarily true without more substantial evidences.

However, to say NVIDIA is selling GTX 460, which is a mainstream product, at loss, you need to provide a better evidence considering NVIDIA is still having a positive gross margin, which certainly make "selling at loss" a smaller possibility.

Also, there is no "proof" here, unless NVIDIA publicly announce that they are actually doing that. And even with it we can't be sure if what you want is "proof." In real world, many decisions are made with mere possibilities, since many things can never be certain. This type of argument comes surprising frequently though. For example, some people say that you don't know whether gravity works the same on other star systems. Of course we don't know, but based on current evidences, we can say that it's very likely to be the same as on earth. So if you want to say that "gravity doesn't work the same on other star systems!" you need to provide a more substantial evidence instead of just saying "you can't never be certain that a star system in such a distance obey to the same physics rules." That's pretty pointless.

Actually, this is ridiculous, if you take one step back.

Bill says that I have evidence that A is true therefore B is false.

Bull. says, look here your evidence is wrong, we still don't know that A is true, that means that we don't know if B is true or false either.

Bill: You're not allowed to say that unless you can show evidence that B is true, the burden of proof is on you since B is less likely to be true. Until then my statement is true and everything else should be ignored

Bull: That's ridiculous.

Bill: I know, but there are special rules in videocards and religion and sex.

Bull: That's cool, I get it ... your d''k is small therefore my d''k is larger...
 
Two words.
Brand recognition.

Not a reason enough, not for performance-midrange that is. People are just going to buy those 460s, cause loses to nV, and won't buy another video card for at least two years (the majority of them, at least). Of course, they will think higher of nV and maybe convince some of their friends to buy some nV profitable cards too.

For my economically challenged brain this makes no sense at all. I do agree that halo products such as 5970 may be (i.e. with higher probability) sold at a loss. Car companies do that some times.

So i suggest finding another reason, if anyone still cares that is.
 
Actually, this is ridiculous, if you take one step back.

Bill says that I have evidence that A is true therefore B is false.

Bull. says, look here your evidence is wrong, we still don't know that A is true, that means that we don't know if B is true or false either.

Bill: You're not allowed to say that unless you can show evidence that B is true, the burden of proof is on you since B is less likely to be true. Until then my statement is true and everything else should be ignored

Bull: That's ridiculous.

Bill: I know, but there are special rules in videocards and religion and sex.

Bull: That's cool, I get it ... your d''k is small therefore my d''k is larger...

Look, did I say that you are not allow to say it's possible that NVIDIA is selling GTX 460 at loss? What I said is that saying it's possible is pointless. Everything is possible. You have to say why it's more likely than not that NVIDIA is selling GTX 460 at loss, otherwise you are just saying something obvious, because many things are possible, although the possibility may be very small.

If you can't understand this, let's me spell it out for you. Do you think AMD is selling most of its GPU at loss? Considering that most of AMD's current GPU are actually smaller than NVIDIA's GPU while having the same performance level and price level, and both have their GPU produced at the same foundry, it's more likely that AMD's production cost is lower. OK? However, the current situation is that AMD is having a much smaller gross margin (not profit) in the GPU sector than NVIDIA's consumer GPU sector. If AMD is not likely to be selling at loss, why should NVIDIA be? So it's less likely for NVIDIA to sell GTX 460 at loss than not.
 
tannat, could you PLEASE change your terminology? A product with positive gross margins but negative operating margins is NOT 'selling at a loss' - every unit is making a profit. It simply did not amortise the R&D/marketing investment over its lifetime. Until everyone insists on using more precise wording, this discussion will go nowhere.

The burden of proof is clearly upon anyone who claims NV or AMD are selling some products with negative gross margins, but arguably it's also upon anyone who thinks every single one of them has paid back its investment. Someone who claims a specific product (e.g. GTX460) did not, however, also has the burden of proof. Obviously, you can always say something is at least possible, but you're not allowed to make ten posts repeating that either *cough*

There, now that we've set the rules of this silly little game, could we stop playing it? :p
 
Look, did I say that you are not allow to say it's possible that NVIDIA is selling GTX 460 at loss? What I said is that saying it's possible is pointless. Everything is possible. You have to say why it's more likely than not that NVIDIA is selling GTX 460 at loss, otherwise you are just saying something obvious, because many things are possible, although the possibility may be very small.

If you can't understand this, let's me spell it out for you. Do you think AMD is selling most of its GPU at loss? Considering that most of AMD's current GPU are actually smaller than NVIDIA's GPU while having the same performance level and price level, and both have their GPU produced at the same foundry, it's more likely that AMD's production cost is lower. OK? However, the current situation is that AMD is having a much smaller gross margin (not profit) in the GPU sector than NVIDIA's consumer GPU sector. If AMD is not likely to be selling at loss, why should NVIDIA be? So it's less likely for NVIDIA to sell GTX 460 at loss than not.


What is likely or not is irrelevant for what I claimed. The only thing I've claimed is that the profit numbers are not a proof for either or neither. I pointed this out since the numbers were used as evidence for one of the outcomes previously.

Is it pointless to point out a false statement just because it points towards the most likely outcome? I point out false statements irrespective of sample and probabilities. As well as I am happy to admit I'm the origin of the wrong statement, as when I mixed up operating profit and manufacturing cost earlier.

Repeatedly truncating sample-size by chopping of the least unlikely scenario will assymptotically lead to wrong answer. Doing it once based on wrong arguments is as flawed, and totally unnecessary.
 
tannat, could you PLEASE change your terminology? A product with positive gross margins but negative operating margins is NOT 'selling at a loss' - every unit is making a profit. It simply did not amortise the R&D/marketing investment over its lifetime. Until everyone insists on using more precise wording, this discussion will go nowhere.

The burden of proof is clearly upon anyone who claims NV or AMD are selling some products with negative gross margins, but arguably it's also upon anyone who thinks every single one of them has paid back its investment. Someone who claims a specific product (e.g. GTX460) did not, however, also has the burden of proof. Obviously, you can always say something is at least possible, but you're not allowed to make ten posts repeating that either *cough*

There, now that we've set the rules of this silly little game, could we stop playing it? :p

Actually, I see that that may be confusing. It's my bad english.
And agreed, I'm so tired of this flamewar.
 
tannat, could you PLEASE change your terminology? A product with positive gross margins but negative operating margins is NOT 'selling at a loss' - every unit is making a profit. It simply did not amortise the R&D/marketing investment over its lifetime. Until everyone insists on using more precise wording, this discussion will go nowhere.
And I would like to mention that negative operating margins are a very difficult thing to estimate, since the R&D on, for instance, the GTX 580 also helps nVidia to build the lower-tier products. So you can't just say that the GTX 580 has negative operating margins when its R&D is also used for other products.

Given that no company has successfully maintained the mid-low end in the discrete GPU market for any significant length of time, my expectation is that the marginal cost in R&D for developing a high-end product instead of just low-mid range products is less than the profit margin on those products.
 
What is likely or not is irrelevant for what I claimed. The only thing I've claimed is that the profit numbers are not a proof for either or neither. I pointed this out since the numbers were used as evidence for one of the outcomes previously.

Simply said, the evidence is to support the argument, not to mathematically proof that. It's impossible to strictly proof anything in the real world, so that's pointless.

Is it pointless to point out a false statement just because it points towards the most likely outcome? I point out false statements irrespective of sample and probabilities. As well as I am happy to admit I'm the origin of the wrong statement, as when I mixed up operating profit and manufacturing cost earlier.

How do you know that's a false statement? You don't know whether NVIDIA is selling anything at loss. It's possible that NVIDIA actually didn't sell any Fermi product at loss. Using a positive gross margin to support this statement may be not strong enough, but it's not necessarily false. A positive gross margin at least doesn't point toward selling at loss, especially when it's better than the company's main competitor. If you want to claim otherwise, just saying "you can't rule out the possibility" is not good enough to persuade others. You need to produce some evidence, such as, maybe NVIDIA had better gross margin before GTX 460 was released, or sales numbers indicating that GTX 460 is not selling that well so 470 and 480's higher margin is big enough to cover the loss, etc.
 
AlphaWolf said:
a boat floats therefore everything on the boat must float

true or false?
Strawman.

One side has hard numbers from a 10K with show GM at or close to their all time high. The other has made up numbers from the local village idiot and his groupies.

Place your bets.
 
And as I point out again, those numbers that are pretty much proof of nothing regarding the profitability of an individual unit out of dozens shown in the filing.
Sure, sure. You're completely making sense.

Be that as it may, which numbers are you bringing to the table as evidence?

(Let's stop this. You're not even trying to come up with arguments. If you had a clue, you would point to inventory write-offs to prove that, at some point in history, they sold some chips for a loss. But you're not doing even that...)
 
Sure, sure. You're completely making sense.

Be that as it may, which numbers are you bringing to the table as evidence?

(Let's stop this. You're not even trying to come up with arguments. If you had a clue, you would point to inventory write-offs to prove that, at some point in history, they sold some chips for a loss. But you're not doing even that...)

Read my posts. Stop trying to attribute claims to me which I have never made. I have never claimed anything other than the possibility exists. I took exception to the claim that there was evidence to show that it was profitable, while such evidence may exist somewhere, it certainly has not been presented in this thread.
 
This seems to be 48SPs 64-Bit GF119: http://www.nvidia.com/object/product-geforce-gt-520m-us.html (significant smaller looking die than on 420M or 525M).

Looking at the pictures, the GT520m package seems to be the same as the GT240m, compare:
http://images.nvidia.com/products/geforce-gt-520m/geforce-gt-520m-med-b.png
and
http://images.nvidia.com/products/geforce_gt_240m/GeForce_gt_240m_back_med.png

The above seems strange as the GT240m and GTS250m which also had the same package were both 128bit chips. Guess this chip needs similar power level to those chips maybe?

If the above info is correct...then package size on the GT240m was 29mm, i think, implying the die size of GT520m is 85mm2 (chip width is 30% of package size and height is 33%)
 
The GT520M has 4 more pins, for what it's worth.

Ahh yeah sorry did miss that...on the left and top on the inside of outermost ring of pins.

How about compare to the GF108/GT540m:
http://images.nvidia.com/products/geforce-gt-540m/geforce-gt-540m-med-b.png

The pin out has the extra 4 pins as for the 520m, though the image is rotated 90degrees.

Comparing GF108 is 36% package width and 37% package height, again using 29mm for package size gives GF108 an area of 114mm2. So new chip is 74% size of GF108.
 
Back
Top