NVIDIA GeForce RTX 4080 SUPER FE Sports Fewer Power Phases Than Non-SUPER Model

[T0@st]

A video review has highlighted some curious internal changes on the NVIDIA GeForce RTX 4080 SUPER Founders Edition graphics card—Geekerwan has discovered that the board design has been "updated" to a small degree, when cross examined with the non-SUPER variant. Team Green engineers have chosen to decrease the number of phases from 13 to 11, while the memory phase count goes from 3 down to 2. HXL (@9550pro) TLDR-ed the situation on social media: "4080 Super FE vs 4080 FE: Core: -2 phase & VRAM: -1 phase." Tech experts have also noticed that the new SUPER FE's board does not have a phase near to its power connector. VideoCardz found these changes to be a little bit odd, considering that the card arrives with faster memory and a increased core count.

Geekerwan reckons that NVIDIA has implemented these internal adjustments in an effort to reduce power consumption in gaming scenarios. The official comparison table confirms this ambition—in the "Average Gaming Power" category we see the GeForce RTX 4080 SUPER target 246 W, and GeForce RTX 4080 (non-SUPER) aim for 251 W. The reviewer notes that their ASUS ROG STRIX RTX 4080 SUPER GAMING sample card features the same power layout as its non-SUPER sibling. They believe that NVIDIA's Founders Edition is the only model bearing an adjusted phase tally—while Team Green's board partners have simply rolled out the previous RTX 4080 layout.




The GPU codename has been revised as well—specifically for review sample purposes. VideoCardz points out: "As noticed by the same reviewer, there is also a small change to the GPU codename, the RTX 4080 SUPER is labeled as PRESS SAMPLE. Typically, reviewers receive early silicon from initial production batches, often labeled as "QUAL SAMPLE" or they have no labels at all. This is the first time NVIDIA has used this new label, presumably only for its Founders Edition. As far as we can tell, this wasn't the thing for RTX 4070 SUPER, while 4070 Ti SUPER and all previous releases (4060 Ti/4060/4070 Ti) did not have a Founders Edition design; hence such a label was not present." The "AD103-400 PRESS SAMPLE GPU" label is visible in this screenshot:



Check out Geekerwan's video analysis on their YouTube channel:

"The last SUPER review of the GeForce RTX 40 series is here. Does the official price reduction mean there is not much improvement? Let's take a brief look at them in this video..."

View at TechPowerUp Main Site | Source

 

[SL2]

Well, isn't that super?

 

[KarymidoN]

waiting for those brazilian guys who add more memory and make all sorts of mods to GPUS to add more phases and memory to make it close to a 4090 for almost half the price.

 

[Philaphlous]

NVIDIA will also be releasing a slightly tuned down version of the "4080 SUPER" called "4080 super" The difference is in the all caps vs non-caps.

 

[Deleted member 229121]

So Nvidia made it cheaper...in every sense of the word.

 

[Enkur]

NVIDIA will also be releasing a slightly tuned down version of the "4080 SUPER" called "4080 super" The difference is in the all caps vs non-caps.

I see what you did there LOL

 

[N/A]

Everything just works so it doesn't matter much. What hurts me the most is they could've used the 4070 super FE board with 12+2 phases pin to pin compatible 256 bit 11 by 17 cm miniitx friendly. Instead of this textolite sheet designed for a 4090.

 

[Hxx]

Nvidia is probably confused - I gave you guys a card that weighs less with less things to break and less things to cool and you guys still complain ? Like wtf

 

[MarsM4N]

Well, why not. Doesn't look like you're loosing any performance. But chance of increased coil whine? Power consumption is at least slightly reduced:

NVIDIA GeForce RTX 4080 Super Founders Edition vs. NVIDIA GeForce RTX 4080 Founders Edition

The board power limit (355W vs. 355W) for overclocking also didn't change.

 

[Random_User]

So Nvidia made it cheaper...in every sense of the word.

Cheaper for them, and their higher margins. And this cut down card still gets the price higher, than older "sturdier" one. Seems like usual nVidia way. And that's after they've already decided to "save" on PCB and components two years ago. As much as during "4080 12GB" fiacso, where they had to shift their entire 4070 price stack.

Maybe they will push saving even more, and replace 12VHPWR/12V-2x6 with older 8 pin as well? Yeah, yeah, 12VHPWR is already there due to cost savings.

 

[SL2]

Cheaper for them, and their higher margins. And this cut down card still gets the price higher, than older "sturdier" one.

Why would anyone buy this card when it's more expensive? It's 1 % faster at 4k.

The whole point was that it should be cheaper, and if it isn't, don't bother. The only pro is gone.

That's in the US, in EU they start at the same price.

 

[trsttte]

This "news" is garbage. Are they comparing the specific components used or just throwing number of phases around? 5W is nothing, card to card variance alone can be higher than. Are they opening up all the cards sold so far? Chances are they did this change months ago in the regular model as well because of anything really, could be a cost optimization or simply component availability for example

Like who gives a shit, phase count was always a marketing point and a bullshit metric, AIBs often use it to try and convince people something is better but it's a meaningless number on it's own.

But chance of increased coil whine?

Or maybe reduced because there are less coils to whine?

It can vary wildly, without further analysing the board design you can't really tell anything about what it could mean

Cheaper for them, and their higher margins. And this cut down card still gets the price higher, than older "sturdier" one.

Why would anyone buy this card when it's more expensive? It's 1 % faster at 4k.

The whole point was that it should be cheaper, and if it isn't, don't bother. The only pro is gone.

That's in the US, in EU they start at the same price.

What are you talking about when the Super version got a 200$ price drop against the original 4080 price!?

Lots of things to be mad about but this ain't it.

 

[SL2]

What are you talking about when the Super version got a 200$ price drop against the original 4080 price!?

What do you think I meant with this? --> "The whole point was that it should be cheaper.."

You've obviously haven't checked the prices, you just blindly follow the MSRP in the reviews. I see one model close to $1000 and in stock, admittedly I didn't see it before. The rest is above 1500.

The prices will of course go down in time, but until then, if the only pro is gone there's no point.

 

[Dr. Dro]

Cheaper for them, and their higher margins. And this cut down card still gets the price higher, than older "sturdier" one. Seems like usual nVidia way. And that's after they've already decided to "save" on PCB and components two years ago. As much as during "4080 12GB" fiacso, where they had to shift their entire 4070 price stack.

Maybe they will push saving even more, and replace 12VHPWR/12V-2x6 with older 8 pin as well? Yeah, yeah, 12VHPWR is already there due to cost savings.

...the 4080 super is cheaper than the original 4080. This is an inconsequential design choice, and if you're doing any extreme overclocking, you'll be running an AIB card with a conventional form factor PCB anyway.

 

[stimpy88]

nGreedia hate you. You were mean and nasty to innocent, perfect nGreedia, so they gave you a gift to calm you down and make you consume more product and complain less.

That gift was less performance in some applications, lower quality cards which might not last so long as more work is spread over less components.

You asked nGreedia for this. Now eat up children.

 

[Dr. Dro]

nGreedia hate you. You were mean and nasty to innocent, perfect nGreedia, so they gave you a gift to calm you down and make you consume more product and complain less.

That gift was less performance in some applications, lower quality cards which might not last so long as more work is spread over less components.

You asked nGreedia for this. Now eat up children.

And this is practically why I have come to resent the AMD fanbase. Can't we all just get along already and acknowledge the fact that neither multi-billion-dollar global megacorporation (almost multi-trillion-dollar, in Nvidia's case) is actually our friends or even remotely care about what the small people think?

 

[MarsM4N]

Or maybe reduced because there are less coils to whine?

It can vary wildly, without further analysing the board design you can't really tell anything about what it could mean

Yea, I was considering that view point, too. But to my logic less coils = more stress on the remaining coils. At least that's the experience as far as I can remember. But unless @W1zzard or someone else does a noise analysis we will never know. Somehow those tests sadly still aren't on the check list of any reviewer, making GPU purchases still a lottery.

Also just compared the reviewed 4080 Super's and the number of power phases do not translate into power consumption. So there are more variables at play that influence the power draw (BIOS, clocks, etc.).

[11 phase][13W pwr idle][302W pwr gaming] NVIDIA GeForce RTX 4080 Super Founders Edition
[11 phase][14W pwr idle][304W pwr gaming] PNY GeForce RTX 4080 Super Verto OC
[12 phase][13W pwr idle][324W pwr gaming] ASUS GeForce RTX 4080 Super TUF OC
[12 phase][20W pwr idle][306W pwr gaming] Palit GeForce RTX 4080 Super GamingPro OC
[13 phase][23W pwr idle][304W pwr gaming] MSI GeForce RTX 4080 Super Expert
[13 phase][28W pwr idle][334W pwr gaming] Zotac GeForce RTX 4080 Super AMP Extreme Airo
[16 phase][24W pwr idle][297W pwr gaming] Galax GeForce RTX 4080 Super SG
[16 phase][16W pwr idle][312W pwr gaming] Gigabyte GeForce RTX 4080 Super Gaming OC
[18 phase][16W pwr idle][315W pwr gaming] ASUS GeForce RTX 4080 Super STRIX OC

 

[trsttte]

Yea, I was considering that view point, too. But to my logic less coils = more stress on the remaining coils. At least that's the experience as far as I can remember. But unless @W1zzard or someone else does a noise analysis we will never know. Somehow those tests sadly still aren't on the check list of any reviewer, making GPU purchases still a lottery.

Also just compared the reviewed 4080 Super's and the number of power phases do not translate into power consumption. So there are more variables at play that influence the power draw (BIOS, clocks, etc.).

[11 phase][13W pwr idle][302W pwr gaming] NVIDIA GeForce RTX 4080 Super Founders Edition
[11 phase][14W pwr idle][304W pwr gaming] PNY GeForce RTX 4080 Super Verto OC
[12 phase][13W pwr idle][324W pwr gaming] ASUS GeForce RTX 4080 Super TUF OC
[12 phase][20W pwr idle][306W pwr gaming] Palit GeForce RTX 4080 Super GamingPro OC
[13 phase][23W pwr idle][304W pwr gaming] MSI GeForce RTX 4080 Super Expert
[13 phase][28W pwr idle][334W pwr gaming] Zotac GeForce RTX 4080 Super AMP Extreme Airo
[16 phase][24W pwr idle][297W pwr gaming] Galax GeForce RTX 4080 Super SG
[16 phase][16W pwr idle][312W pwr gaming] Gigabyte GeForce RTX 4080 Super Gaming OC
[18 phase][16W pwr idle][315W pwr gaming] ASUS GeForce RTX 4080 Super STRIX OC

Exactly, it can vary wildly, a lot of stuff goes into the design of these things. Less coils can be more stress on the remaining coils if they're using the same components, or less if they bought higher capacity ones. Higher number of phases can also mean they're using the cheapest crapiest components so they need a lot of them to meet their power target, or they found a nice deal bulk buying the same components used in cheaper cards. Or maybe some guy from the marketing department came to the engineering floor and asked them to go nuts with a very high number of phases because higher number = better in marketing materials (that asus strix oc and gigabyte oc come to mind)

 

[TheBeastInside]

200$ less and as far as anyone so far can tell, no real impact on anything else.
Yes, they're saving money to make more money, but isn't that what companies do?
It's not like the initial scare people had with the 3080 / 3090s.

 

[pavle]

I would be surprised if that whole board (with components) cost nvidia more than $5. Cheap shyte.

 

[Dr. Dro]

I would be surprised if that whole board (with components) cost nvidia more than $5. Cheap shyte.

It does cost them more than $5, although a lot less than $1000. What you're forgetting though, is the cost of developing software for ten years plus to support your one-time purchase, and the research and development costs of creating this product to begin with. Neither of which are cheap. They aren't a charity, after all, and after considering all of that, they still must turn out a profit.

This has been reflecting in AMD's GPU business, they have not hesitated to discontinue hardware that they don't feel like spending any further engineering resources on regardless of the product's age or initial cost, we've seen this with the R9 Fury X and the Radeon VII, and in the Vega Frontier Edition's case, they literally just stopped releasing any sort of targeted software maintenance to it far before declaring EOL. If you want to use it for gaming, better know your way around the registry, otherwise, it'll run their Pro drivers but at the same time, support practically none of their Pro features either, unless you flash a WX 9100 BIOS onto it. Fun!

 

[Random_User]

Spoiler

I would be surprised if that whole board (with components) cost nvidia more than $5. Cheap shyte.

Not the first time. Remember, how Nvidia cards were imbued with GPUs falling off PCB, due to crappy BGA solder. The same stuff had hit 4870X2 I had about the same time.

Spoiler

It does cost them more than $5, although a lot less than $1000. What you're forgetting though, is the cost of developing software for ten years plus to support your one-time purchase, and the research and development costs of creating this product to begin with. Neither of which are cheap. They aren't a charity, after all, and after considering all of that, they still must turn out a profit.

This has been reflecting in AMD's GPU business, they have not hesitated to discontinue hardware that they don't feel like spending any further engineering resources on regardless of the product's age or initial cost, we've seen this with the R9 Fury X and the Radeon VII, and in the Vega Frontier Edition's case, they literally just stopped releasing any sort of targeted software maintenance to it far before declaring EOL. If you want to use it for gaming, better know your way around the registry, otherwise, it'll run their Pro drivers but at the same time, support practically none of their Pro features either, unless you flash a WX 9100 BIOS onto it. Fun!

It's funny, since you've yourself previously mentioned, that the corporations are not consumer's friends, but yet find so many reasons to justify their actions and wrongdoings. No offence, though. But R&D is expensive. This isn't news. It was always pricy, even in years of less refined nodes, like 28nm or bigger. It was as expensive for companies as it is now, with 4-5nm, because at that times, they had a lot less money, and budgets were many times smaller. So the top nodes at that times hit them the same way.
And I'm pretty sure, that the foundry's premium tax, is much smaller than the one Nvidia slaps on their products (and AMD as well). If it was that pricey, there woudn't steady profit margin growth, that never drop below 50%.
Something remains the same. Another one is a habit to cheap out on crucial components and VRAM. There's a lot of other stuff they can save on, but they deliberately decided to reduce the amount of phases. But who cares, after gullible people would buy this stuff. This move might still be fine for this card, but every time such thing happens, it drops the confidence even more.
You've said that these companies are not charities. But based on what happens now, it looks as they are some charity foundations, where masses of uninformed consumers give them tremendous amount of money, for things that in fact const several times less. How this isn't charitable organizations?

 

[Dr. Dro]

Spoiler

Not the first time. Remember, how Nvidia cards were imbued with GPUs falling off PCB, due to crappy BGA solder. The same stuff had hit 4870X2 I had about the same time.

Not this again. This bumpgate nonsense is literally 17-18 years in the past. How is this even remotely close to bumpgate?

 

[Totally]

Why would anyone buy this card when it's more expensive? It's 1 % faster at 4k.

The whole point was that it should be cheaper, and if it isn't, don't bother. The only pro is gone.

That's in the US, in EU they start at the same price.

Supers have never been cheaper prior to this launch, last launch shifted every tier to the next highest price bracket and people still bought them by the pallet.

 

[Noyand]

Spoiler

It's funny, since you've yourself previously mentioned, that the corporations are not consumer's friends, but yet find so many reasons to justify their actions and wrongdoings. No offence, though. But R&D is expensive. This isn't news. It was always pricy, even in years of less refined nodes, like 28nm or bigger. It was as expensive for companies as it is now, with 4-5nm, because at that times, they had a lot less money, and budgets were many times smaller. So the top nodes at that times hit them the same way.

Yes, they have much more money, but also but much more areas that they're doing business in. NVIDIA product stack in 2024, isn't comparable to what they had in 2011. And TSMC also stopped doing volume discount.

TSMC Ends Its Volume Discounts For the Biggest Customers, Could Drive Product Prices Up

Taiwan Semiconductor Manufacturing Company (TSMC), one of the largest semiconductor manufacturers in the world, is reportedly ending its volume discounts. The company is the maker of the currently smallest manufacturing nodes, like 7 nm and 5 nm. For its biggest customers, TSMC used to offer a...

www.techpowerup.com