NVIDIA Updates GeForce RTX 4080 Silicon with AD103-301 SKU

[ARF]

You'd be surprised.

Personally I think the current GPU prices will result in a big drop in PC gaming in general, which we will start to see in a year or two.

The average person cant spend £800 on a GPU.

I think it will take more than a year or two but we will see PC gaming level off instead of the continued rising of the past. Probably the next step will be for gamers to realize that you don't need ultra settings and just turn the settings down a little and keep using the card that you already have.

In my case I can easily afford a $1,000 card but I won't be ripped off.

I also can throw a grand to the problem but I think neither company deserves to get such a significant donation from me.
They are all our no-friends, so why would we want to support their greedy schemes, even if we are enthusiasts.

I think I will buy something at the moment when at first sight I recognise it as a deal.
Examples - Radeon HD 4890 for $195, or Radeon R9 380 for €230.

 

[TheinsanegamerN]

There is no balance in extremely well cooled RTX 4080 designs because they all adhere to previously used RTX 4090 size form factor.
The users are getting robbed of the ability to have more case compatible designs. Thermal headroom is not much of an issue. Same story for RTX 4070 Ti designs.

Anyone who is sure that AIB's have no other abilities to design cards differently (in size, shape) has been properly programmed by NVIDIA to think so. Their ingenuity has to go through some strict guidelines before they are able to come out to market. This is very much NVIDIA's iron fist.

View attachment 279380
This is why we don't have things like this tiny ZOTAC GTX 1080 Ti anymore, not until AIBs are allowed to do so again

So why do AIBs have PCBs larger then nvidias, those would also have to be approved no? Do you have a source for nvidia denying small PCBs?

 

[Dirt Chip]

Well.... a $1 BOM savings is like getting excited about saving .02 on a pack of bubblegum because you had a coupon, hehehe

I wonder how much an NV 'RT on' bubblegum will cost...

 

[bug]

My appologies I did not really make it clear but I was more so meaning local PC shops in general will be hurt. We are a statewide shop and one of the only service centers in the area so we can compete with some retailers but on big ticket items like GPUs and CPUs it is nearly impossible. I do know this will hurt other stores in our area though who have some Ada stock (if they do a big price cut).

Also Nvidia does not do price drop credits like apple, amd, and intel. So if they undercut you a week after you order a box of cards you will still be invoiced for full price.

No apology needed, but it all makes more sense now.

It's not only stores, apparently EVGA got out of the whole business because of disagreements with Nvidia's policy. Hopefully there's a way to make them clean up their act like that. Fingers crossed lower-than-expected sales is a first step towards that.

And when you say "price drop credits", does AMD cover that if some AIB drops their prices? Or only if they lower the GPU prices themselves? Or is that something AIBs do, but Nvidia won't let them?

 

[mama]

Tests will need to be done to verify the claims about no loss of performance.

 

[dj-electric]

There won't be any technical capability to cool a 400-watt RTX 4090 with that "tiny" thing.
There are small liquid cooled RTX 4090, though, and if you are smart, these are the only viable options to buy.

View attachment 279393

MSI GeForce RTX™ 4090 SUPRIM LIQUID X 24G

While I mentioned RTX 4080 and RTX 4070 Ti cards specifically in regards to design freedom, I would even include the RTX 4090 in my argument.
Have you seen how overengineered their designs are? Here, i'll help

Some of those figures make this seem like a 200W card, not a 450W+ one. There is so much thermal headroom to think of shrinking the RTX 4090 in size. Thanks to the big core surface and vapor chamber designs companies could really go wild, but they don't because they can't. Not yet at least. Designs go through NVIDIA's review before coming out to market. My sources? this is NVIDIA, i'm not going to make a shooting gallery out of my sources. (@TheinsanegamerN)

Not just the RTX 4080 and RTX 4070 Ti can enjoy smaller form factors, so can the RTX 4090. Yes, even with its 450W+ power envelope. Nobody can convince me RTX 4090 cards have to be this large, and a very nice and large table from ours truly confirms it. There are 15, 20, 25 degree Celsius figures here in distance from any remote trouble, not to account the fact most of those fan RPMs barely whisper. The thermal headroom for cooler design shrinking is unreal.

 

[N/A]

What headroom. 60-66 is perfectly normal temperature. This is not a CPU to be reaching 99 like it's perfectly safe and widely accepted.

 

[Hxx]

I think it will take more than a year or two but we will see PC gaming level off instead of the continued rising of the past. Probably the next step will be for gamers to realize that you don't need ultra settings and just turn the settings down a little and keep using the card that you already have.

In my case I can easily afford a $1,000 card but I won't be ripped off.

due to the chip shortages and high silicon demand we;ve seen these companies adapting and getting smarter. inventory production can be ramped up or down (we are seeing this now with 4090s and 80s) and none of these brands would need to worry about providing price cuts on their most expensive silicon. I think that's the trend we are going to see... less sales and discounts and a more controlled inventory volume. I think the consumer mentality will also change from "wait for a sale" to "save money buy as high end as u can and hold out as long as possible".

 

[bug]

While I mentioned RTX 4080 and RTX 4070 Ti cards specifically in regards to design freedom, I would even include the RTX 4090 in my argument.
Have you seen how overengineered their designs are? Here, i'll help
View attachment 279483
Some of those figures make this seem like a 200W card, not a 450W+ one. There is so much thermal headroom to think of shrinking the RTX 4090 in size. Thanks to the big core surface and vapor chamber designs companies could really go wild, but they don't because they can't. Not yet at least. Designs go through NVIDIA's review before coming out to market. My sources? this is NVIDIA, i'm not going to make a shooting gallery out of my sources. (@TheinsanegamerN)

Not just the RTX 4080 and RTX 4070 Ti can enjoy smaller form factors, so can the RTX 4090. Yes, even with its 450W+ power envelope. Nobody can convince me RTX 4090 cards have to be this large, and a very nice and large table from ours truly confirms it. There are 15, 20, 25 degree Celsius figures here in distance from any remote trouble, not to account the fact most of those fan RPMs barely whisper. The thermal headroom for cooler design shrinking is unreal.

I'm too lazy to link every picture here, but PCB shots in each review show pretty different designs. I'm not seeing Nvidia's stranglehold. Unless you count placing VRMs next to each other, in a straight line maybe.
Or maybe you mean freedom in choosing a heatsink. Which AIBs have been done since forever and I have never seen anything about changes in that area.

 

[Minus Infinity]

Does $1 BoM = $400 retail, because $799 is where the 4080 needs to be.

 

[dj-electric]

I'm too lazy to link every picture here, but PCB shots in each review show pretty different designs. I'm not seeing Nvidia's stranglehold. Unless you count placing VRMs next to each other, in a straight line maybe.
Or maybe you mean freedom in choosing a heatsink. Which AIBs have been done since forever and I have never seen anything about changes in that area.

Size and form factor. That's all I've been about so far.

 

[bug]

Size and form factor. That's all I've been about so far.

And you know Nvidia enforces those, but you can't tell us how you know that. Right?

 

[dj-electric]

And you know Nvidia enforces those, but you can't tell us how you know that. Right?

Sort of, yeah. My word is only the word of some random person in some forum, so im not expecting anyone to take it particularly seriously.
That said - I am hoping that sales figures of RTX 40 cards will change their minds and let in some flexibility in design, stuff that will allow RTX 4080 cards look more like RTX 4070 Ti cards and not copycats of RTX 4090 cards (which they often very much are, duplicating designs over). RTX 4080\90 cards that are only up to or lower than 3 slots are possible. RTX 4080\90 cards that are shorter than 32cm and only have 2 fans are possible, RTX 4080\90 cards that may have 3 fans but aren't 15cm in height are possible. I'm waiting and hoping to ever see those

 

[bug]

Sort of, yeah. My word is only the word of some random person in some forum, so im not expecting anyone to take it particularly seriously.
That said - I am hoping that sales figures of RTX 40 cards will change their minds and let in some flexibility in design, stuff that will allow RTX 4080 cards look more like RTX 4070 Ti cards and not copycats of RTX 4090 cards (which they often very much are, duplicating designs over)

I'm with you that cards have been looking ridiculous for quite some time. I even prompted W1zzard to add a con in reviews for that (spoiler alert: he won't).
But as another random in some forum, I have not one single shred of evidence Nvidia or AMD dictates card sizes.

 

[ARF]

What headroom. 60-66 is perfectly normal temperature. This is not a CPU to be reaching 99 like it's perfectly safe and widely accepted.

60-66 is dead cold. There is at least 20 degrees overhead left on the table, instead the user must torture themselves how to fit the disgusting thing in their case.

 

[dj-electric]

60-66 is dead cold. There is at least 20 degrees overhead left on the table, instead the user must torture themselves how to fit the disgusting thing in their case.

Not just that, the fan system in most models works quite effortlessly in dissipating heat. 950-1300RPM we see on those coolers is extremely conservative and very quiet during full loads. The thermal headroom is truly huge if you want to make a more compact card. It will run hotter, and possibly a little louder, but entirely within spec and reasonable accoustics.

Now imagine that. RTX 4080 coolers are copycats of RTX 4090 models, and thermal headroom is even larger, much larger. Yet we still got RTX 4080 models identical in form factor to the RTX 4090 cards they are based on. These things truly don't fit in so many cases on the market. The RTX 4080 is one of the first times in doing this whole thing were I legitimately was disappointed by how excessively large these cards are for the amount of power they take. Seeing users having trouble fitting them in many regular ATX cases isn't making this much better.

We used to have options, we used to pick if we want a compact card that might run louder and hotter, just so we can fit it in smaller enclosures. Now our selection in high end GPUs goes from just "very long" at about 32cm to "how will this fit anywhere?!" at over 36cm in some models. Not to speak of the terrible combo of card height with the combination of a power adapter where you get to points even cases like a Meshify 2 won't close on without applying awkward force on the side panel. And all of that for what? for excessively low operating temeperatures on most models, even when the fans barely spin.

 

[AdmiralThrawn]

No apology needed, but it all makes more sense now.

It's not only stores, apparently EVGA got out of the whole business because of disagreements with Nvidia's policy. Hopefully there's a way to make them clean up their act like that. Fingers crossed lower-than-expected sales is a first step towards that.

And when you say "price drop credits", does AMD cover that if some AIB drops their prices? Or only if they lower the GPU prices themselves? Or is that something AIBs do, but Nvidia won't let them?

If you are an intel Platinum partner or an apple dealer and they lower the msrp of something you just ordered stock of. They will send you a check for the difference so you do not lose money. NVIDIA has never done this as far as I know.

 

[bug]

If you are an intel Platinum partner or an apple dealer and they lower the msrp of something you just ordered stock of. They will send you a check for the difference so you do not lose money. NVIDIA has never done this as far as I know.

Intel and Apple are different, they sell their own products. Nvidia and AMD have to go through AIBs, that's why I asked specifically for how AMD handles this.

 

[PenguinBelly]

Yes, I remember graphics chips having the ability to even work with different types of memory, but those times are long behind us. As far as I know NVIDIA is exclusively selling GPU+memory bundles, and given the level of integration and dependency between those components I'm not surprised. NVIDIA isn't offering alternative memory configuration on on their own consumer cards either (the last generation being Kepler if memory serves). For professional cards the only recent one I can think of is A100 being available in 40 and 80GB versions. So yes, memory configuration is dictated by NVIDIA.

You talk like you know how graphics cards are built today (compared to yesterday). How are they different?

 

[80251]

@PenguinBelly I seem to remember that there was a low tier Nvidia card that worked with different types of VRAM, but that was four generations ago, before Pascal and might have even been before Maxwell.

 

[bug]

You talk like you know how graphics cards are built today (compared to yesterday). How are they different?

Precisely the way you'd expect, from an engineering point of view.

When GPUs were simpler, tolerances were higher. Now that GPUs are incredibly complex, tolerances had to be tightened. For example, in the days of Radeon 8500 you were able to choose between 64 or 120MB VRAM. And the smart choice was to go for less VRAM, because that was lower latency. Such liberties in designing video cards are long gone.
And let's not forget transistors got physically smaller and frequencies went way up. That means much more tight voltage and timings control.

 

[ARF]

Not just that, the fan system in most models works quite effortlessly in dissipating heat. 950-1300RPM we see on those coolers is extremely conservative and very quiet during full loads. The thermal headroom is truly huge if you want to make a more compact card. It will run hotter, and possibly a little louder, but entirely within spec and reasonable accoustics.

Now imagine that. RTX 4080 coolers are copycats of RTX 4090 models, and thermal headroom is even larger, much larger. Yet we still got RTX 4080 models identical in form factor to the RTX 4090 cards they are based on. These things truly don't fit in so many cases on the market. The RTX 4080 is one of the first times in doing this whole thing were I legitimately was disappointed by how excessively large these cards are for the amount of power they take. Seeing users having trouble fitting them in many regular ATX cases isn't making this much better.

We used to have options, we used to pick if we want a compact card that might run louder and hotter, just so we can fit it in smaller enclosures. Now our selection in high end GPUs goes from just "very long" at about 32cm to "how will this fit anywhere?!" at over 36cm in some models. Not to speak of the terrible combo of card height with the combination of a power adapter where you get to points even cases like a Meshify 2 won't close on without applying awkward force on the side panel. And all of that for what? for excessively low operating temeperatures on most models, even when the fans barely spin.

The thermal capacity of that giant cooler with six fat heat pipes is equal to an ordinary liquid cooler.
Which leads to the question - why didn't they use compact liquid coolers for all FE and AIB cards instead of throwing so much metal at the problem? Are they fighting some type of contracts with the metal producers? And what about the ecology - is it sustainable to waste so much material - so instead of making 5 small cards, they now make 1 large?

 

[Mussels]

They wont do this to give people a better card, but to either cut costs or to add a safety feature to prevent the power cables melting - something on the GPU side to sense if things are off with the connector and shut off

 

[gffermari]

…and give the AIBs a chance to compete with the FEs.

 

[dj-electric]

Which leads to the question - why didn't they use compact liquid coolers for all FE and AIB cards instead of throwing so much metal at the problem? Are they fighting some type of contracts with the metal producers? And what about the ecology - is it sustainable to waste so much material - so instead of making 5 small cards, they now make 1 large?

Well, quite frankly liquid coolers are ticking time bombs, especially if blessed by Asetek's domestic quality production. I consider anything with a water pump a ticking time bomb. You can always have an easier time fixing up and refurbishing an air cooler fan system. I'm not huge on the idea of WC as a must on cards. All I really want is more compact and just a genuine variety of form factors for high end cards.

One of the problems that now arise from the use of these gigantic cooler designs is that companies like MSI, ASUS and GBT started slightly altering and using them on Radeon cards too, so those now also fit into the monster size category for a lot of models. Thankfully AMD does not force particular form factors and design guidelines on its AIBs.