• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Discussion Ada/'Lovelace'? Next gen Nvidia gaming architecture speculation

Page 113 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Hmm, not sure why people would want that if it uses more power and is physically larger. I doubt there is much of a performance difference but it will be interesting to see if it behaves differently WRT frequencies.
 

4070 Ti Super with AD102 dies? That's very cut down.
I wonder if this particular variant, assuming it costs the same as other 4070 Ti Supers, will OC better since it has more silicon area to dissipate heat.
 
Does anyone else have issues with Ada series GPUs using DSC at 4K240Hz?

I have a few problems that I attributed to the cable but persist even with a few others that didn't have problems without DSC at 160Hz.
Namely a bit of flicker when the GPU is at a low power and e.g. the wallpaper changes.

Edit: I realize this may not be the right thread... my bad. I created a thread here https://forums.anandtech.com/threads/rtx-4090-dsc-issues.2620304/
 

GB202 12*8 512-bit GDDR7 = 192 SMs
GB203 7*6 256-bit GDDR7 = 84 SMs
GB205 5*5 192-bit GDDR7 = 50 SMs
GB206 3*6 128-bit GDDR7 = 36 SMs
GB207 2*5 128-bit GDDR6 = 20 SMs

GB207 is smaller than AD107. Coupled with N4P, we'll see a 100mm² GPU die? Return of xx50 class (<=$199) GPUs to Desktop?
Hahaha, no way. Whatever you think the xx50 class card will be, that’s the new xx60 card. Adjust expectations accordingly 😂

Late edit: They've had options in that price range pretty much all along. 1650 flavors, 1660, and right now the 3050 6 GB & 8 GB variants. I just don't expect anything that starts with a series number of "5" to be under $200 any time soon, if there is a xx50 in the next, what, year(?) I'd be really surprised.
 
Last edited:
GB207 is smaller than AD107. Coupled with N4P, we'll see a 100mm² GPU die? Return of xx50 class (<=$199) GPUs to Desktop?

I'm assuming it would mainly be for notebooks and probably low-end Quadro cards since anyone using the Turing-based ones is probably looking for an upgrade soon. That's going to give them a much better ASP than the consumer desktop GPU market, even if they charge $300.

Perhaps it'll eventually be used for a xx50 refresh like GA107 was, as the initial 3050 was a cut down GA106 die. Then again AD107 was used for the 4060, so we may not even see a 5050 outside of mobile.

Edit: Removed misplaced formatting tag.
 
Last edited:

nVidia is at it again with the de-branded SKUs, this time the "3050 A Laptop" which is an Ada model.

Presumably it has less than the 20 SMs that the 4050 laptop does and perhaps some other cuts.
 

nVidia is at it again with the de-branded SKUs, this time the "3050 A Laptop" which is an Ada model.

Presumably it has less than the 20 SMs that the 4050 laptop does and perhaps some other cuts.
What a weird thing.

And I doubt this will be cheap at all.
 
One thing I don't understand is why its based on AD106?

RTX 4060 and 4050 are based on AD107. 3050 A will be AD106? Broken dies that Nvidia will later switch to AD 107?
 

The 4070 with GDDR6's speed is 20 gbps. Specs otherwise are the same. That's 5% less bandwidth so it should be slower but only so much.
 
Are they just going to manufacture this variant going forward? It seems unusual to build something like this at all while keeping the older models in production.
 
I would assume so. Lines up with the idea that Blackwell does not support GDDR6X at all.

I'm sure that GDDR6X makes no financial sense anymore when GDDR7 is available, since 6X uses the expensive PAM4 signalling, while GDDR7 uses the more economical PAM3.

Furthermore, Nvidia surely pays a premium for GDDR6X since they are wedded to a single supplier for it, while they can play suppliers against each other for GDDR7.
 
It’s concerning how many companies, especially in the tech gadget industry, prioritize stock value over customer satisfaction, leading to declining service and ethics. Despite high customer dissatisfaction, stock values stay up due to short-term financial metrics, monopolistic practices, or lack of competition. However, this approach is unsustainable—neglecting customers risks long-term brand loyalty and growth. Companies must balance financial goals with genuine customer care to thrive, especially in the fast-evolving tech gadget market.
Write a haiku about Nvidia prioritizing stock value and short term financials over customer satisfaction with regards to their RTX GPU's.
 
Back
Top