Speculation and Rumors: Nvidia Blackwell ...

The only thing bad about the RTX 4080 was its launch price. The performance gain over the 3080 was enough to be considered a true successor, and complaining about the performance gap to the 4090 doesn't make any sense. It's clear that the 4090 represented an entirely new class of graphics card, a super-flagship sitting far and above AMD's 7900XTX and Nvidia's previous flagships in performance and transistor count and priced accordingly. And the 3080 was a fluke, it was far closer to the 3090 than Nvidia because it used the same GA102 die, when Nvidia likely originally intended it to be on the 103 die like the 4080. Prior generations don't have a xx90 to serve as a comparison anyways. If the 5080 is the same price or cheaper than the 4080 Super and delivers a significant performance increase then that would be great regardless of what the 5090 is.
 
Nvidia doesn't eliminate anything, they add on top of what would be the top end card otherwise and to do this they introduce a new product at a higher price. If not for 2080Ti then 2080 would be the fastest Turing, 4080 would be the top card of Lovelace and 5080 would have probably been the top card of Blackwell. The cards above that are an extension of the previous pricing range (for a single card anyway) to produce an SKU for those who can afford it. Anyone thinking that a 4090 could have been sold for $700 at launch are delusional.
 
raytracingfan said:
The only thing bad about the RTX 4080 was its launch price. The performance gain over the 3080 was enough to be considered a true successor
Click to expand...
Its launch price was directly related to its naming, though. The 4080 was basically the equivalent of the 3070 in Ampere terms - a cut down upper midrange die. Now I'm not saying we should have expected the 4080 to cost $500 like the 3070, but it shows how incredibly they exploited naming here to upcharge consumers, on top of the further upcharge even at the same naming tier. Had the 4080 been called a 4070Ti for $750, I think plenty of people would have been quite alright with it, while still giving Nvidia an effective $250(or 50%) price hike. Wouldn't have been an amazing situation, but a decent enough 'meet in the middle' between consumer and corporate wants.

If we're talking the performance gain, we aren't paying for performance, we're paying for the hardware. Pascal gave us huge performance gains without charging us out the nose for it. Pascal is beloved, but it obviously wouldn't have been if they'd charged $900 for the GTX1070. That's essentially what Lovelace did.

I expect the 5080 to provide a fairly modest increase in performance on the 4080, given the rumored specs, and to again try the same $1200 price point, while positioning it as the 'high end' for this generation, even though it's really not. This technically lets them compare the value of the original 4080's price rather than its current price to make it seem like a better deal, even though they're still selling us like a 3070Ti-level GPU for $1200. Then perhaps six months from now or something, they can reduce it back to $1000 and slot in a 5080Ti above it.
 
Seanspeed said:
I expect the 5080 to provide a fairly modest increase in performance on the 4080, given the rumored specs, and to again try the same $1200 price point, while positioning it as the 'high end' for this generation, even though it's really not. This technically lets them compare the value of the original 4080's price rather than its current price to make it seem like a better deal, even though they're still selling us like a 3070Ti-level GPU for $1200. Then perhaps six months from now or something, they can reduce it back to $1000 and slot in a 5080Ti above it.
Click to expand...
The 4080 Super launching at $1K was effectively an admission from Nvidia that the original 4080 was priced too high. I doubt they will go for $1200 again, but there's a chance they'll get greedy since the RDNA4 flagship will be weaker than the 5080.
 
MOD MODE: I'm quickly tiring of the pendanticism. If you can't make your point without being obviously obtuse or combative, then don't post. The topic of "what makes a tier" is a set theory problem, not a technical one. You're welcome to talk about how you define your pariticular set, however you're not allowed to brow-beat someone about how their definition isn't right.

Next overtly snarky reply gets nuked. More than one in a row gets someone a several-day break from posting.
 
Now, back to normal contributor mode, I agree with @pjbliverpool in expressing sadness to the forum's ability to continually try to hash this out. Big long passionate opinions on why any particular vendor (NVIDIA, AMD, Intel) named their parts they way they did, why somehow it's still wrong, and why other people who disagree are wrong.

In the end, bluntly, it doesn't fucking matter. NVIDIA could call the next "almost-but-not-quite-toppest-of-the-line" card the IXBQ#7 and that's now the name and we just get to buy it or not.

Here's where reason and logic take over: consumers who are buying this part are doing so either because of name recognition (NVIDIA, RTX, the "x080 series", whatever) or based on performance figures from reviews and benchmarks on social media. And if one customer who buys solely on the name recognition of the "x080-series" nameplate is sorely let down with their purchase, then they're at least moderately likely to remember the expensive lesson during their next GPU purchase.
 
raytracingfan said:
complaining about the performance gap to the 4090 doesn't make any sense.
Click to expand...

Agreed, the value prop for the x80 card isn’t based on how close it gets to the x90. The x90 can be $4000 and 4x faster but that doesn’t matter to someone with a $1000 budget.

The expectation that prices and relative performance are anchored to some fixed ratios every generation is unrealistic as these things can change at any time.
 
raytracingfan said:
The 4080 Super launching at $1K was effectively an admission from Nvidia that the original 4080 was priced too high. I doubt they will go for $1200 again, but there's a chance they'll get greedy since the RDNA4 flagship will be weaker than the 5080.
Click to expand...
The only lesson Nvidia has learned is that they need to provide the utmost minimal improvements in performance per dollar for consumers and fanboy morons to defend them.

People praised the 'Super' series value for Lovelace, even though the original value was just so garbage that any slight improvement got positive attention.

Nvidia has learned that 99.99% of consumers are uninformed morons and easily exploitable. I've long been an advocate of the idea that the biggest problem with corporate greed is not their inherent and predictable greed, but the consumer weakness in giving in because they dont want to do without. AKA - consumers lack principles.
 
Seanspeed said:
The only lesson Nvidia has learned is that they need to provide the utmost minimal improvements in performance per dollar for consumers and fanboy morons to defend them.

People praised the 'Super' series value for Lovelace, even though the original value was just so garbage that any slight improvement got positive attention.

Nvidia has learned that 99.99% of consumers are uninformed morons and easily exploitable. I've long been an advocate of the idea that the biggest problem with corporate greed is not their inherent and predictable greed, but the consumer weakness in giving in because they dont want to do without. AKA - consumers lack principles.
Click to expand...
This seems pretty pessimistic to me.

If you want the highest performing consumer video card available today, you're buying from NVIDIA. If you want the second highest performing consumer video available today, you're still buying from NVIDIA. If you want to talk about "bang for the buck", then NVIDIA may still very well be in the mix depending on the performance target you're aiming for.

I can conceive of no good argument in saying "consumers lack principles" when it's unreasonable to ask them to stop paying for top tier performance. Who are you to say they cannot or should not spend the money they want in buying those items? Who are any of us to say NVIDIA shouldn't charge what the market will bear? This is capitalism 101. Ferrari makes the SF90 not because everyone can afford to buy them, but because some people in this world want the fastest, sexiest, Ferrari-reddest thing on the road. Why shouldn't Ferrari sell as many as they can reasonably make, or at least as many as they care to?

Do I wish video cards were less expensive? You bet. Do I have at least two dozen options to spend less money to get less performance? Yes, I absolutely do.

Consumers have a LOT of choices in this space today. Apparently, you don't agree with the choices they're making; that doesn't make your opinion any more right than theirs.
 
Last edited:
Kopetite has been wrong before - both the bus width and SM count are higher than expected on the 5090. It seems to work against their effort to push AI users to the workstation cards, unless they’ve managed to gimp the cards for AI (although in that case would they really need D variants for the 5080?)
 
Albuquerque said:
In the end, bluntly, it doesn't fucking matter.
Click to expand...
Which is exactly my point however I am being told that some mythical "x80 tier" means something for a product which hasn't come out yet and thus the only thing we know about it is it's supposed marketing name.

There are tiers: performance tiers, pricing tiers, power consumption tiers, cooler size tiers, etc. I.e. tiers of actually important and comparable characteristics between products of one or more vendors.

There are no "naming tiers". Product names don't mean anything. Even indication of a generation and a relative positioning inside this generation product lines aren't necessarily a part of marketing naming. Nvidia (and other vendors) are abusing this glitch in consumer understanding to "upsell" them products which aren't necessarily better than a differently named product - people are stuck in this mindset where a product is somehow different just because it's a "40 series" or "x80 SKU" when in practice the only thing which should matter to a consumer are performance/cost/features.

And I truly hope that we can end this discussion on that and never ever mention a "x80 tier" in a TECHNICAL DISCUSSION ever again.

Edit: And here's another wrench into the whole "but names mean something to me" discussion:

 
Last edited:
You must log in or register to reply here.
Back
Top