NVidia Ada Speculation, Rumours and Discussion

[DegustatoR]

No, more like 16-23%.

1080p without RT is irrelevant for anything above 3060 already so that 16% means about nothing.
Then there's RT in which case 3090Ti can be up to 35% faster than 3080.
So I stand by my figure of ~30% faster (25%-35%). If GA104 based 4080 will end up being above that mark while costing the same as 3080 MSRP then it will be a fine option really, if not very attractive to the owners of 3080 but it is rare that you have a reason to upgrade at the same price level each generation.

And that's with the 3090Ti already being pushed much harder in terms of power limits and whatnot as well, it's not just the raw spec difference which really isn't that big.

Not sure why this matters. 4090 will be comparable to 3090Ti in how it will be pushed.

We're supposed to get significant improvements in performance per dollar each generation.

We are? Where's the rule which says this?

What we actually are supposed to do is to decide for ourselves if the new products worth the asking price against those which we have. How these products are named and which chips they are using is completely irrelevant.

 

[Jawed]

Yay we've reached the point in the new release cycle where people who only buy NVidia are desperately hoping AMD will compete hard enough...

 

[trinibwoy]

Depends on what you mean by 'in the dust'.

And no, it isn't all that matters. We're supposed to get significant improvements in performance per dollar each generation. Not a creeping 10-20% more, but like 40%+ per pricing tier.

I dont know why y'all would be ok with this.

No one said they would be ok with 10-20% more performance. Why are you assuming that it won’t be 40%+?

If the 4080 12GB is 40% faster than the 3080 and the 4080 16GB is 30% faster than the 4080 12GB and the 4090 is 50% faster than the 4080 16GB would the 4080 12GB still be considered an overpriced POS because it’s not the bestest?

Either way there is no guarantee of 40% improvement in performance per dollar. Those days are over with the slowdown in manufacturing process improvements and increase in cost. It also depends on what performance you’re referring to. RT at 4K will certainly see big gains if that’s your cup of tea.

 

[DavidGraham]

leading to its relatively poor sales.

What poor sales? Turing dominated the market and marketshare numbers.

 

[techuse]

People will still lineup to buy Nvidia GPUs regardless of how outlandishly they are priced.

 

[Flappy Pannus]

People will still lineup to buy Nvidia GPUs regardless of how outlandishly they are priced.

No, they won't. PC gaming is not an unlimited market.

 

[DegustatoR]

People will still lineup to buy Nvidia GPUs regardless of how outlandishly they are priced.

That has more to do with the fact that there are no alternatives which would be better products at lower prices than "people" or "Nvidia".

 

[OlegSH]

Yay we've reached the point in the new release cycle where people who only buy NVidia are desperately hoping AMD will compete hard enough...

I guess competing hard enough would be launching new products first at lower prices rather than following up competitors pricing, yet here we are.

 

[Seanspeed]

1080p without RT is irrelevant for anything above 3060 already so that 16% means about nothing.

That 16% figure was for 1440p. 23% for 4k. I deliberately left out the 1080p results(which would have only been 11%) because I was trying to be reasonable.

You can 'stand by' your claims all you want, but it's not correct. Using some outlier examples is not an honest representation of its typical capabilities.

As for why the 3090Ti using higher power levels matters, it's because the comparison was to a standard 3080. The point being the gap in capabilities really wasn't that big, as you were trying to claim.

It's disappointing to see such entitlement in a technical forum.

It's disappointing to think that just because it's a technical forum means we cant remember that we're all still consumers, and that getting yanked around by corporate greed is not cool.

 

[Seanspeed]

Either way there is no guarantee of 40% improvement in performance per dollar.

I mean, that's up to Nvidia. These new GPU's will represent a huge uplift in performance on a like for like basis(die size). The only reason we wouldn't get at least a 40% increase in performance per dollar is if Nvidia decides to raise prices and deliberately delivers worse value for customers because they think they can get away with it.

I really dont understand why y'all are trying to act like such an expectation is so unreasonable. :/

 

[trinibwoy]

I mean, that's up to Nvidia. These new GPU's will represent a huge uplift in performance on a like for like basis(die size). The only reason we wouldn't get at least a 40% increase in performance per dollar is if Nvidia decides to raise prices and deliberately delivers worse value for customers because they think they can get away with it.

I really dont understand why y'all are trying to act like such an expectation is so unreasonable. :/

What like for like basis? A 5nm die costs more than an 8nm die of the same size. Of course we all want big performance increases at lower prices but acting like there’s some magic number we’re entitled to is just silly.

 

[eastmen]

It's disappointing to see such entitlement in a technical forum.

What is the point of newer hardware if it doesn't perform better than previous ? Seems like entitlement from the gpu makers to me. Everyone on the forum can make a personal choice of how little performance increase they are willing to accept

 

[DegustatoR]

What is the point of newer hardware if it doesn't perform better than previous ? Seems like entitlement from the gpu makers to me. Everyone on the forum can make a personal choice of how little performance increase they are willing to accept

Which h/w doesn't perform better than previous?

 

[eastmen]

Which h/w doesn't perform better than previous?

were you unable to read my 3 sentences in their entirety ?

 

[DegustatoR]

were you unable to read my 3 sentences in their entirety ?

Nowhere in your "three sentences" you describe what h/w you are talking about. Please do.

 

[Below2D]

It's disappointing to see such entitlement in a technical forum.

Expecting better products is entitlement now?

 

[pjbliverpool]

Which h/w doesn't perform better than previous?

I think he was referring to an earlier post talking about performance per dollar. Or more specifically "significant improvements in performance per dollar each generation". Personally I think that's a pretty reasonable expectation and wouldn't consider it entitled. More like it's one of the key drivers that sustains the PC enthusiast market. If we only get performance improvements at ever higher prices, or have to wait years for very small improvements at the same price then I don't see the PC enthusiast market as being sustainable, unfortunately.

For example if the 4080 12GB offers only a 30% performance uplift over the 3080 and also increases price by $100 then that's worse than the Pascal to Turing transition which was generally considered disappointing at the time in performance terms. But at least in that case Turing brought along all the RTX features (which where vastly underappreciated at the time of launch). Even at the same price, after a 2 year wait that wouldn't be over compelling.

 

[DegustatoR]

I think he was referring to an earlier post talking about performance per dollar. Or more specifically "significant improvements in performance per dollar each generation". Personally I think that's a pretty reasonable expectation and wouldn't consider it entitled. More like it's one of the key drivers that sustains the PC enthusiast market. If we only get performance improvements at ever higher prices, or have to wait years for very small improvements at the same price then I don't see the PC enthusiast market as being sustainable, unfortunately.

For example if the 4080 12GB offers only a 30% performance uplift over the 3080 and also increases price by $100 then that's worse than the Pascal to Turing transition which was generally considered disappointing at the time in performance terms. But at least in that case Turing brought along all the RTX features (which where vastly underappreciated at the time of launch). Even at the same price, after a 2 year wait that wouldn't be over compelling.

I don't agree with any of that.

A. "Significant" isn't a measurable definition. +30% in CPU ST is significant, right? Why is it not in GPU performance? Then what are we using when we count this number? Average performance results from the likes of HUB who still pretty much ignore RT in 2022? TPU results do not include RT in them either so what is the baseline for such comparisons? Do I personally care if some new card provide "only" +30% in 1080p in Diablo 3? Is this "insignificant"?

B. +30% at the same price DOES improve perf/price. Nowhere anywhere does anyone promised you that it will be more (or less) with each next generation. If you don't like it then don't buy the new product, it's really that simple.

C. Pascal to Turing was more or less flat on perf/price (1080Ti was slightly behind 2080 at the same launch price). This isn't the same as a +30% gain in performance at the same price.

Then what products are we even talking about? Have I missed the announcement of MSRPs for the upcoming cards? Do we know what else they will bring to the market besides the oh so popular "moar FLOPs"? People are too quick to jump at any possibility to bash some product before they even know anything about it - price including. Remember the times when 4080 was expected to draw 500W and everyone and their mother was running around screaming how unacceptable that is? What happened to this I wonder.

 

[DavidGraham]

NVIDIA cryptically confirmed the announcement of next gen, confirmed the Ada name of next series, confirmed the die size (629mm) and number of transistors as well (75.38 billion).

NVIDIA confirms upcoming Ada architecture - VideoCardz.com

NVIDIA shares a teaser for the upcoming Ada architecture NVIDIA released a new teaser for its “GeForce Beyond” broadcast next week. It has now confirmed the name upcoming architecture. NVIDIA teases Ada Architecture, Source: NVIDIA The Twitter teaser posted a “Diagram for the computation by the...

videocardz.com

 

[neckthrough]

I was specifically talking about the language “We’re supposed to”. That kind of language devalues the struggles that thousands of engineers in material sciences, process technology, circuits, architecture and algorithms are facing in continuing the pace of exponential scaling that the whole world has become accustomed to.

What other field has shown such sustained exponential progress? Cars? Houses? Airplanes? Dams? Food?

And now it’s just HARD.

So no, we’re not “supposed to” get anything. We *get* what’s feasible within the constraints of technology and economics. And as consumers we buy or we don’t and get the final say in whether that market segment is sustainable or not.