Nvidia shows signs in [2023]

  • Thread starter Deleted member 2197
  • Start date
Status
Not open for further replies.
Yeah, the card is probably the worst in perf/price gains over Ampere gen.
No, it's pretty much as bad as anything else in the Lovelace lineup. It's a consistently ridiculously terrible lineup of GPU's, all doing the same thing of selling lower end GPU's at higher end prices and trying to legitimize it with their renamings.

Almost every Lovelace GPU has a ~100% price increase from the Ampere generation in terms of what you're actually getting.
AD106 will probably end up being about 24B of Ts? This puts it closer to GA102's 28.3B than GA104's 17.4B. Do any of these dies look "low end" to anyone?
Again, I know you're not dumb enough to believe what you're saying here. 'Cost per transistor' isn't an actual thing. Costs are based on wafers. And there's a significant difference in the amount of usable dies you get from a smaller die compared to a larger die, which is why small dies are used in low end products and large dies are used in high end products. Vast difference in overall yields.

You know all this perfectly well, of course.
 
Again, I know you're not dumb enough to believe what you're saying here.
I know what I'm saying here and it's not me who's dumb to not listen and resort to constant personal attacks instead of providing any sort of relevancy to the discussion.

'Cost per transistor' isn't an actual thing.
Yes, it is.

Costs are based on wafers.
Are you selling wafers in your GPUs?
Is the cost of a wafer a constant set by law?

And there's a significant difference in the amount of usable dies you get from a smaller die compared to a larger die, which is why small dies are used in low end products and large dies are used in high end products. Vast difference in overall yields.
Yields don't tell us anything about a cost of some chip made on some wafer either. You have to know the cost of the wafer and the yield to know how much a chip of some size would cost to make on that wafer.

You know all this perfectly well, of course.
I know perfectly well that you made a 100% false claim when you've said that a chip is "low end" because it's ~200mm^2.

The 4060 would have been a nice 4050 for $150.
Why not $50 or $5?
 
Nice try. Are you going to defend this inconsistency by pretending the chip size was unknown?
I'm not going to defend anything. The chip size doesn't tell us anything about it market position.

Since there seem to be many people who don't understand it - wafer cost is not a fixed sum and change rather dramatically between processes and over time.
 
I know perfectly well that you made a 100% false claim when you've said that a chip is "low end" because it's ~200mm^2.
To believe that a sub 200mm² GPU in the modern market isn't a 'low end' part would really require supreme idiocy. And I dont for a second think you're actually that dumb.

You're just dishonest to the bone. You will literally say anything to defend Nvidia.
 
Come the frick on dude. Who are you trying to fool here? What is wrong with you?
Again, I know you're not dumb enough to believe what you're saying here
You're just dishonest to the bone. You will literally say anything to defend Nvidia.
I know you're not dumb enough to actually believe that
Please stop your personal attacks and concentrate on the arguments at hand. Your behavior is not conducive to a proper discussion and doesn't actually help your case, stick to the facts in a civilized manner.
 
I'm not going to defend anything. The chip size doesn't tell us anything about it market position.
Amazing plan.
Chip size telling us nothing about the market position would mean chip cost does not affect the market position. Yet historically chip sizes corresponded to market positions. We never saw bigger chips made for lower segments or vice versa. Can you explain that?
 
Come the frick on dude. Who are you trying to fool here? What is wrong with you?
Again, I know you're not dumb enough to believe what you're saying here
You're just dishonest to the bone. You will literally say anything to defend Nvidia.
I know you're not dumb enough to actually believe that
Please stop your personal attacks and concentrate on the arguments at hand. Your behavior is not conducive to a proper discussion and doesn't actually help your case, stick to the facts in a civilized manner.
Indeed. I haven't time to clean up this thread but any future personal attacks will see thread bans.

One issue I see with this conversation is people having different definitions. If you spent more time spelling out what exactly you're saying instead of insulting people for not seeing what your seeing, you might get a better conversation.

Take the time to explain your understanding of chips costs, wafer cost per mm², manufacturing costs, and product positioning so everyone knows eactly what they are disagreeing with!
 
"Historically" doesn't prove anything for the future. Things change.


You haven't seen, say, Vega 10 competing with GP104?
Do you really go through life refusing all inductive knowledge?
And how could such a change occur?

I did see them competing. But to stay on topic, never did I saw for example GP104 in $150 or $1500 products.
 
Do you really go through life refusing all inductive knowledge?
And how could such a change occur?
Oh, I don't know. Maybe it has something to do with the realities of silicon production?

I did see them competing. But to stay on topic, never did I saw for example GP104 in $150 or $1500 products.
So? Now you see AD103 in a $1200 product. What does that tell you?
Also since you haven't seen it apparently here's a product using GP104 for $3000.
 
Can we please just establish the points rather than relative comparisons and telling everyone rhey're dumb for not seeing things your way?

1) What exactly is the argument? 200 mm² doesn't constitute a high end GPU? What's the evidence for and against? What counts as a high-end GPU? Performance or pricing?
2) For pricing, what are the factors being looked at as affecting? Can we point to different dies of the same sort of size at considerably different prices? If talking about wafers, are we talking about the same wafers or different wafers for different nodes?

These one-line back-and-forths aren't meaningful discussion. If the discussion is worth having, take a breath and have a proper go at it. If it's not worth discussing properly, just walk away before a frustrated mod who hasn't time to clean up threads and doesn't want a forum/thread-wide nuke just starts dishing out thread bans without really following the discussion to see who's 'right'.
 
Status
Not open for further replies.
Back
Top