Discussion NVIDIA "Ada Lovelace" Flagship GPU rumored to feature 18176 Cores, 48 GB "24 Gbps" GDDR6X MEM & 800W TBP !

  • Thread starter Thread starter Deleted member 2731765
  • Start date Start date
Toggle sidebar Toggle sidebar
D

Deleted member 2731765

Guest
Hey guys,

More leaks and rumors on Nvidia's upcoming RTX 40-series GPU lineup. But before continuing, I have one question.

Do gamers really need an 800-900 Watt and 48GB GPU, just to game on the extreme 4K Max in-game settings ? Well, I suppose the answer depends on how many are willing to shell money on a "luxury" or a high-end product, be it a GPU or any other gadget. And, not to mention the high power consumption and heat/output.

But anyways, some new details regarding the upcoming flagship NVIDIA "Ada Lovelace" GPU SKU have once again been reported by Kopite7kimi, which makes it seem like the ultimate graphics chip ever made. I mean flagship amongst flagships.

This isn't the first time that such a high-end Ada Lovelace GPU SKU has been talked about though. The previous rumor, which also came from the same leaker, reported on a Titan-class graphics card within the Ada Lovelace GPU lineup which would feature some insane specifications. Once again, this isn't the fully enabled AD102 GPU, while the previous variant was mentioned as a 900W TBP SKU with the full 18432 CUDA Cores.

According to the rumored specs, the graphics card will be rocking the Ada Lovelace architecture & feature a slightly cut-down configuration "AD102-450-A1", rocking 142 SMs (out of 144 SMs) on 18,176 CUDA cores (out of 18432 CUDA cores). Based on a rumored clock speed of around 3 GHz, this graphics card could easily break past the 100 TFLOPs compute barrier.

The graphics card is said to be equipped with 48 GB of GDDR6X memory running across a 384-bit bus interface. What's interesting is that NVIDIA will not hold back on the VRAM specs and feature the latest 24 Gbps memory modules, delivering up to 1.152 TB/s of VRAM bandwidth to the GPU. That's a 14% increase in memory bandwidth compared to the existing RTX 3090 Ti flagship which features 21 Gbps memory dies.

The upcoming RTX 4090 is also expected to utilize the same 21 Gbps memory dies, with only the flagship 'Ti' model getting the 24 Gbps dies. As for power consumption, the new NVIDIA Flagship AD102 GPU-powered graphics card will be insane, with almost double the TDP of the RTX 3090 Ti, rated at up to 800W.

Considering that a single 16-pin connector can only provide 600 Watts of power, a dual 16-pin connector config will have to be utilized for this monster of a card, IF it ever becomes a reality. The graphics card may utilize the PG137-SKU0 board.

I have a feeling this might be a workstation GPU though, instead of gaming. Based on the Ampere lineup, we saw that NVIDIA not only didn't release a Titan graphics card, but essentially replaced the Titan series with its BFGPU-class GeForce RTX lineup. The higher capacity cards still launched as the workstation RTX A**** lineup which also got the full GA102 treatment, but besides the RTX 3090 Ti and the RTX A6000, there was no Titan class SKU.

So will a Titan-class GPU make sense for Ada Lovelace or will this specific SKU end up as the next-gen gaming BFGPU and/or workstation flagship? Well, we can't say for sure, but one thing is definitely sure that such a GPU configuration will indeed be insane in terms of specifications, power draw, and price.

The card, if it ever comes to retail, will definitely launch after the RTX 4090 which is expected to be unveiled later this fall, and will obviosuly cost a fortune ! But I digress. :sweatsmile:

https://twitter.com/x/status/1551619750572457984
View: https://twitter.com/kopite7kimi/status/1551619750572457984

https://twitter.com/x/status/1532341054246354944
View: https://twitter.com/kopite7kimi/status/1532341054246354944
 
  • Like
Reactions: Roland Of Gilead and alceryes
Sort by date Sort by votes
Seeing the next-gen flagship GPUs duking it out will definitely be something to behold!! I can't wait to microwave some popcorn and watch the show.
I really hope AMD makes huge strides in the hardware ray-tracing, FSR, and streaming/recording performance segments, as they are a bit lacking.

48GBs of RAM tells me that this card is targeting the productivity workstation audience. For gaming, they could've cut the amount of memory in half (24GBs) and wouldn't lose one bit of performance, even with the next 3-4 years of AAA games (guessing here).
 
  • Like
Reactions: Roland Of Gilead, PEnns and Deleted member 2731765
Upvote 1 Downvote
alceryes said:
48GBs of RAM tells me that this card is targeting the productivity workstation audience. For gaming, they could've cut the amount of memory in half (24GBs) and wouldn't lose one bit of performance, even with the next 3-4 years of AAA games (guessing here).
Click to expand...

Yeah, most definitely. A 48GB VRAM GPU would be pointless for gaming, even on 4K resolution, with highest texture/aliasing settings, and also with RTX/RAY Tracing enabled. Maybe for 8K resolution this much extra GPU frame buffer might help, but who knows what the future holds, lol. 8K res is a FAR CRY though....

But like you have also mentioned, 24 GB should be more than enough for playing the latest AAA graphic demanding games on 4K resolution, along with some extra eye candy settings, for few more years.

I'm actually more worried about the power consumption and PRICE of these new upcoming GPUs. Because the power consumption has also increased for most of the mainstream and mid-range GPUs, along with the price.

A 48GB GDDR6X VRAM card on the other hand, would surely be extremely expensive, but if it is for the workstation segment, then the price would be justified IMO.
 
  • Like
Reactions: alceryes and Dark Lord of Tech
Upvote 1 Downvote
Metal Messiah. said:
I'm actually more worried about the power consumption and PRICE of these new upcoming GPUs. Because the power consumption has also increased for most of the mainstream and mid-range GPUs, along with the price.
Click to expand...
800W is insane. Especially if it's anything like previous gen cards where they don't take any OC or power limit increase into account.
I REALLY hope NVIDIA got the transient power spikes under control or there's gonna be a whole host of people who think their PSU is enough only to find that it's unstable at random (heavy load) times. There may be a run on 1000W+ PSUs later this year due to next gen cards. The TDP increase for the Ryzen 7000 CPUs is one reason why I upgraded from my perfectly good 750W RMX to my SuperNOVA P6 1000W. It was on sale for $160 - couldn't pass it up.
 
  • Like
Reactions: Deleted member 2731765
Upvote 0 Downvote
The xx60 class GPUs won't see a very huge bump in VRAM as well as TDP though.

alceryes said:
The TDP increase for the Ryzen 7000 CPUs is one reason why I upgraded from my perfectly good 750W RMX to my SuperNOVA P6 1000W.
Click to expand...

Are 7000 CPU series that much power hungry ? I thought AMD has optimized the efficiency ratio by a fair margin in ZEN 4 arch ?
 
Upvote 0 Downvote
Metal Messiah. said:
The xx60 class GPUs won't see a very huge bump in VRAM as well as TDP though.
Are 7000 CPU series that much power hungry ? I thought AMD has optimized the efficiency ratio by a fair margin in ZEN 4 arch ?
Click to expand...
The AM5 platform received a pretty big TDP and PPT spec increase over AM4. In the neighborhood of 35%. It's been rumored that a 'certain SKU' in the Ryzen 7000 CPU family will have a TDP of 170W. I enjoy chasing benchmark numbers so I will certainly push my new Ryzen 7000 CPU. Remember, the AM5 platform will, supposedly, have a long life. Maybe the first gen Ryzen 7000 won't need it but two generations down the road I'll probably be glad I picked up a 1000W PSU.
 
  • Like
Reactions: Deleted member 2731765
Upvote 0 Downvote
spentshells said:
Very exciting, but it leaves me wondering about a 306-3060 class card

8GB of ram 2560 or rather >4000 shadder, Im interested to see how efficient these cards can be.
Click to expand...
The 4070-class cards may see a VRAM increase if all the 4080-class (and above) cards go up to 12+GBs. Maybe the 4070 will be rocking 10GBs...?
I don't see it happening for the 4060 cards though. The 8GB VRAM limit will be there to remind people that they bought a mid-level performance card. While rasterization power may be good enough, you'll need to bump texture level and other VRAM hungry eye-candy settings down a bit to make sure everything can fit in the 8GBs of VRAM with new AAA games over the next few years.
OR
They'll do the same thing they did with the RTX 3060 (non Ti). Increase the amount of VRAM (strictly for marketing reasons) and eunuch the bus width.
 
  • Like
Reactions: spentshells and Deleted member 2731765
Upvote 0 Downvote
Believe it or not, I'm not so worried about 800 Watt cards. Who remembers 390X Crossfire rigs and how much they consumed? What I'm more worried about is the cooler that will be required to cool these things. These will be true enthusiast class alone. The dimensions will be enormous. I wouldn't be surprised to see a 4 fan card that is extra long and takes up 3+ slots. That or high flow rate liquid will be mandatory. I can't imagine a vertical tower board with an air cooler that beefy, unless you have a GPU riser support. I'll be happy with something that works with my 850Watt PSU
 
  • Like
Reactions: alceryes and Deleted member 2731765
Upvote 0 Downvote
Well, it appears these high-end GPUs are on the horizon. And sort of confirmed as well, the SKU model names. The Eurasian Economic Commission or EEC has just registered several NVIDIA GeForce RTX 40 "Ada Lovelace" and AMD Radeon RX 7000 "RDNA 3" graphics cards.

Although the EEC is just a registry where trademarks are assigned when a company submits their upcoming products, the resource has been very useful as most of the trademarked products end up launching as retail products sooner or later. We have seen several products in the past that got registered and made it to the retail segment such as graphics cards from AMD, NVIDIA & their AIB partners. This latest registration spree comes from AIB AFOX.

Btw, I've never heard of this AFOX GPU brand/AIB before though.

FYmgJX-akAMMLH8.png


FYmgJX6agAAmDVy.png


The products include:

NVIDIA GeForce RTX 40 "Ada Lovelace" Series:
  • GeForce RTX 4090 Ti
  • GeForce RTX 4090
  • GeForce RTX 4080 Ti
  • GeForce RTX 4080
  • GeForce RTX 4070 Ti
  • GeForce RTX 4070
  • GeForce RTX 4060 Ti
  • GeForce RTX 4060
  • GeForce RTX 4050 Ti
  • GeForce RTX 4050
The NVIDIA GeForce RTX 40 series lineup seems to be made up of both "Ti" and "Non-Ti" models for each GPU segment. There are reports that the RTX 4090 (Non-Ti) might end up as the only 40 series cards to launch this year while the rest of the lineup is expected in early Q1 2023.

AMD Radeon RX 7000 "RDNA 3" Series:
  • Radeon RX 7900 XT
  • Radeon RX 7900
  • Radeon RX 7800 XT
  • Radeon RX 7800
  • Radeon RX 7700 XT
  • Radeon RX 7700
  • Radeon RX 7600 XT
  • Radeon RX 7600
  • Radeon RX 7500 XT
  • Radeon RX 7500
MOST importantly, and a bit surprising as well, the above list also mentions NVIDIA's GeForce RTX 30 SUPER series lineup, but we know that the green team isn't preparing any new Ampere cards since it already has an overstock of existing GPUs in its inventory to get rid of and it is likely that the "SUPER" lineup was replaced by the "Ti" offerings and soft-refreshes such as the RTX 3080 12 GB graphics card.

So an RTX 30 SUPER SKU seems highly unlikely.

Source:

https://twitter.com/x/status/1551948012888666112
View: https://twitter.com/harukaze5719/status/1551948012888666112
 
Last edited by a moderator:
  • Like
Reactions: alceryes
Upvote 0 Downvote
Metal Messiah. said:
BTW, It looks like you won't have to wait too long before you can grab a new Ryzen 7000 CPU. AMD might possibly launch these SKUs on or before the 4'th of AUGUST, 2022.

AMD Ryzen 7000-Series Likely to Launch On or Before the 4th of August | TechPowerUp
Click to expand...
Nice! Didn't think they were releasing that early. (still speculation, but...)

Something I'm thinking about is how much of a savings we'll see on BF/CM vs initial retail release. If it's in the range of hundreds of $$$ I'd be pissed that I bought at release. This may be why they're releasing a bit early. More of a chance of people paying the initial MSRP instead of waiting 3.5 months for a possible sales. Then we have the scarcity factor. Retailers now know that they can charge whatever the Hell they want when product is scarce and no one can do anything about it.

Knowing me however, I'll be tempted by a nice Microcenter bundle and drop $1.5k+ on the CPU + MB + DDR 5 + NVMe SSD + case + CPU coolong solution. 😀
 
Last edited:
Upvote 0 Downvote
digitalgriffin said:
Believe it or not, I'm not so worried about 800 Watt cards. Who remembers 390X Crossfire rigs and how much they consumed? What I'm more worried about is the cooler that will be required to cool these things. These will be true enthusiast class alone. The dimensions will be enormous. I wouldn't be surprised to see a 4 fan card that is extra long and takes up 3+ slots. That or high flow rate liquid will be mandatory. I can't imagine a vertical tower board with an air cooler that beefy, unless you have a GPU riser support. I'll be happy with something that works with my 850Watt PSU
Click to expand...
They can't make the cards any longer without making them unable to fit in most standard mid or full tower cases so a 4-fan card is unlikely. 3 slots thick may become the norm for flagship cards though.
 
Last edited:
Upvote 0 Downvote
alceryes said:
They can't make the cards any longer without making them unable to fit in most standard mid or full tower cases so a 4-fan card is unlikely. 3 slots thick may become the norm for flagship cards though.
Click to expand...

You are quite correct, there is little room for expansion here. But I don't think a triple slot will handle that much power. Four slots will be too heavy. That leaves with liquid 360, 280, 240 solutions. At 240, that will be a heck of a thick radiator by my calcs.
 
Upvote 0 Downvote
digitalgriffin said:
You are quite correct, there is little room for expansion here. But I don't think a triple slot will handle that much power. Four slots will be too heavy. That leaves with liquid 360, 280, 240 solutions. At 240, that will be a heck of a thick radiator by my calcs.
Click to expand...
Agreed.

The 4090 Ti is in a thermal solution class all its own. I'm guessing closed loop 240, minimum.
The non-Ti 4090 and 4080s will probably be right on the cusp of what a good 3-slot air cooler can maintain. They don't need to maintain peak draw cooling 100% of the time but something like ~350W average with spikes to 420W is doable (or whatever combined metric they go by). Those fans are going to be loud though!
Let's not forget about GPU sag too! The half-life of these air-cooled solutions will be short as the sag causes gaps to appear over time in heatsink/plate contact, worsening an already tenuous cooling solution.
 
  • Like
Reactions: Deleted member 2731765
Upvote 0 Downvote
What's this rumor all about ? RTX 40 series coming so soon, less than a month's time ? Most likely this won't turn out to be true.

The rumor comes from Ilya Korneychuk of PRO-Hi-Tech who recently stated in his YouTube video that the first batch of NVIDIA's upcoming GeForce RTX 40 series graphics cards is expected to be dispatched in less than a month. The YouTuber alleges that the information comes from his own Chinese Supplier but this shouldn't mean that the cards are expected to hit retail shelves the same day, as the embargo on that may be very different.

So a supply of the next-gen graphics cards going out within a month's timeframe seems highly unlikely.

PRO Hi-Tech: Chinese GPU supplier to receive NVIDIA GeForce RTX 40 cards in less than a month - VideoCardz.com

Twitter-Ilya-RTX40-768x585.jpg
 
Last edited by a moderator:
Upvote 0 Downvote
Some update.

According to a new rumor published over at Chiphell Forums, the NVIDIA GeForce RTX 4090 graphics card will come in both 450W & 600W variants & be priced lower than the GeForce RTX 3090 Ti's MSRP. So more SKUs of the same card model I suppose ?

The rumor states that the NVIDIA GeForce RTX 4090 might be the only "Ada Lovelace" powered graphics card launching and hitting retail shelves this year with the other variants possibly slipping into Q1 2023.

The more interesting part is the TGP figures. It is stated that the NVIDIA GeForce RTX 4090 graphics card will come in both 450W and 600W TGP flavors. The reference design is most likely going to be kept at 450W but AIBs will have the option to go up to 600W in their custom designs.

Furthermore, it is clarified that the 450W and 600W variants will have the same frequency (boost clock) limit but the higher TGP variant will have a lower chance of hitting its TGP wall and may end up providing a slightly better performance uplift within games. NVIDIA's AIB partners will mention if the card uses the 450W or 600W TGP figure on their box packages which is something that we haven't seen in the last few generations.

Finally, and the most important bit of this rumor, the Forum member states that NVIDIA's GeForce RTX 4090 will have a lower MSRP than the RTX 3090 Ti. The RTX 3090 Ti had an MSRP of $1999 US & the RTX 3090 had an MSRP of $1499 US.
 
  • Like
Reactions: alceryes
Upvote 0 Downvote
YUP, 4 USD less ! Haha, On a serious note it's kind of hard to predict the market price these days, at least for the gaming GPU segment. I just hope another covid attack or lockdown doesn't destroy the whole supply/demand chain like it did 2 years ago.
 
Upvote 0 Downvote
Some more updates and rumors on the RTX 40 series. It looks like there is also going to be a Ti variant of the AD104 SKU.

According to leaker Kopite7kimi, the NVIDIA GeForce RTX 4070 Ti might end up being a beast of a card, offering performance that matches the GeForce RTX 3090 Ti, the existing flagship. The leaker reported on some new specifications and it looks like the 4070 series will be benefiting from much higher power limits than their predecessors.

The NVIDIA GeForce RTX 4070 Ti is expected to utilize the full-fat AD104 GPU core with 7680 cores or 60 SM units. From previous leaks, we also know that the AD104 GPU will come packed with 48 MB of L2 cache and up to 160 ROPs which is simply insane. That's a 25% increase in core count and a 12x increase in cache versus the GeForce RTX 3070 Ti which rocks the GA104 GPU core.

As for memory specs, the GeForce RTX 4070 Ti is expected to rock 12 GB GDDR6X capacities that are said to be clocked at 21 Gbps speeds across a 192-bit bus interface for 504 GB/s bandwidth. While the RTX 4070 is expected to rock a TBP of 300W, the RTX 4070 Ti can end up with a power limit as high as 400W. Now, this could be the power limit for custom designs with the Founders Edition roaming around the 325-350W power limit. The graphics card is expected to be based on the PG141-SKU331 PCB design

https://twitter.com/x/status/1553925364954853376
View: https://twitter.com/kopite7kimi/status/1553925364954853376
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom