NVIDIA Updates GeForce RTX 4080 Silicon with AD103-301 SKU

[AleksandarK]

NVIDIA has reportedly begun shipping NVIDIA GeForce RTX 4080 graphics cards with a newer GPU SKU that changes the requirement for PCB design and is set to lower manufacturing costs. Previously, the company shipped its AD103-300-A1 SKU to power the GeForce RTX 4080 graphics cards. However, the new AD103-301 SKU will power the upcoming RTX 4080 cards that the company plans to ship to its AIBs and possibly use in the reference design. With the new 301 version, the GPU performance and power envelope should not change. What does change is the PCB design requirements, as the new SKU revision possesses a different chip pinout that doesn't correspond to the old design.

HKEPC has reported that GPUs with AD103-301 SKU are shipping, while VideoCardz confirms the AIB update with Gainward also offering updated cards. GALAX offers RTX 4080 models with either AD103-300/301 as well. Additionally, the GeForce RTX 4070 Ti will also see an SKU update, with AD104-250 being replaced by AD104-251. With these new silicon revisions, customers will not see any difference. However, the AIBs and NVIDIA could see a cost reduction to improve margins. HKEPC estimates around $1 BOM cost reduction with the new SKU, which will make a difference in thousands of cards shipped.



View at TechPowerUp Main Site | Source

 

[qubit]

I hope it's also that bit more power efficient and allows better overclocking.

 

[dj-electric]

You know what NVIDIA should really do?
Open up PCB design and complete card dimentions and cooler designs for AIB's free will. They should stop gatekeeping designs behind size and shape demands and just let them get creative.
Those who wonder why graphics cards look the certain way, and are very similar to each other, that's why.

 

[ZoneDymo]

I hope it's also that bit more power efficient and allows better overclocking.

Better is always better but tbf the 4080 is a very impressive card when it comes to efficiency

 

[ncrs]

You know what NVIDIA should really do?
Open up PCB design and complete card dimentions and cooler designs for AIB's free will. They should stop gatekeeping designs behind size and shape demands and just let them get creative.
Those who wonder why graphics cards look the certain way, and are very similar to each other, that's why.

What makes you think they aren't open to AIB's free will? Looking at the latest 4080 reviews I can see that the NVIDIA's PCB is distinct from PNY's, which is in turn distinct from MSI's and ASUS'.

 

[Vayra86]

Cheaper manufacturing... surely they aren't creating the narrative for a price drop here? Card's only just out.

I know, I'm dreaming.

 

[the54thvoid]

Cheaper manufacturing... surely they aren't creating the narrative for a price drop here? Card's only just out.

I know, I'm dreaming.

Sales are poor. NV and AMD have miscalculated consumer enthusiasm for $1+k goods. You never know; leak this as a face saving exercise in humility, allows a price drop without admitting they got it wrong?

Though a one dollar decrease in BOM. How much can that equate to, really?

 

[bug]

Sales are poor. NV and AMD have miscalculated consumer enthusiasm for $1+k goods. You never know; leak this as a face saving exercise in humility, allows a price drop without admitting they got it wrong?

Though a one dollar decrease in BOM. How much can that equate to, really?

Like the article says, that's more of a cost saving for those making thousands and tens of thousands of PCBs.

 

[xorbe]

This cost savings is not for us. It's for the middlemen between nv and us.

 

[dj-electric]

What makes you think they aren't open to AIB's free will? Looking at the latest 4080 reviews I can see that the NVIDIA's PCB is distinct from PNY's, which is in turn distinct from MSI's and ASUS'.

This is not free will, this was done under a guardian eye. Free will is take that graphics chip and do what you want with it without having NVIDIA halting your design.
This is why we don't have compact RTX 4080 designs.

Among other things - that will also mean different memory configs, capacity. Remember that? Now, I know that part might be asking too much, but asking is also free

 

[bug]

This is not free will, this was done under a guardian eye. Free will is take that graphics chip and do what you want with it without having NVIDIA halting your design.
This is why we don't have compact RTX 4080 designs.

Among other things - that will also mean different memory configs, capacity. Remember that? Now, I know that part might be asking too much, but asking is also free

You're conveniently leaving something out: Nvidia still has to provide support for said chips. If they simply let AIB go wild, that would dramatically increase support costs. And you'd still end up paying for it, make no mistake.

This isn't about choosing "free will" (really?) and an iron fist, it's simply about striking a balance. And as usual, when you have to strike a balance, there will always be some that will feel cheated.

Since this isn't something we can influence, I'm really not worrying about it.

 

[lightning70]

If there is an improvement in costs, do you think there will be a decrease in the price of the product?
Will the RTX 4080 be $1,050 or $1,000?
Will the consumer receive a discount?

 

[N3utro]

There is a cost to changing the design of the cards to this new PCB. I doubt it's worth it for just a 1$ difference. There is probably another reason behind this.

Let's just hope that this doesn't mean the new PCBs are of lower quality than the first ones and that nvidia is trying to lure buyers with it after all the 4080 reviews have already been done...

 

[ncrs]

This is not free will, this was done under a guardian eye.

As in following a specification? For example, the chip is meant to work with a certain level of power delivery, but the specifics are left to AIBs - and from my examples earlier you can see that each has a slightly different power delivery system with different number of phases, different devices and even capacitor types. I doubt NVIDIA dictates all of this with an iron fist, but not everything is as flexible.

Free will is take that graphics chip and do what you want with it without having NVIDIA halting your design.
This is why we don't have compact RTX 4080 designs.

The chip is specified for a certain level of heat emission with standard power limits. Making a compact 4080 would mean taking compromises like lower performance due to decreased power limits. I am unsure whether any AIB vendor would like to risk having a niche product like that.
There's also the, somewhat irrational, "bigger is better" mentality in this. The bigger card must be better, right?
An alternative cooler design was recently reported here on TPU as well, so that doesn't seem to be dictated by NVIDIA either.

Among other things - that will also mean different memory configs, capacity. Remember that? Now, I know that part might be asking too much, but asking is also free

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.

 

[Crackong]

$1 BOM reduction

AIB: Is that mean I could have a $1 discount?
Nvidia: .....

 

[Space Lynx]

I think this has more to do with the fact that people were making fun of Nvidia for placing the power connection so close to hot components on the PCB... I saw it in a couple of videos, even one being an engineer in the field, saying how dumb of a PCB design it was.

I wouldn't be surprised if the new cards we see move the connector elsewhere.

 

[bug]

I think this has more to do with the fact that people were making fun of Nvidia for placing the power connection so close to hot components on the PCB... I saw it in a couple of videos, even one being an engineer in the field, saying how dumb of a PCB design it was.

I wouldn't be surprised if the new cards we see move the connector elsewhere.

And respinning the silicon is the most straightforward solution for a misplaced power connector. Right?

If they shuffled pins around, the most likely target was simpler tracing on the PCB. Best case scenario (but unlikely), manufacturers can get rid of one PCB layer.

 

[Space Lynx]

And respinning the silicon is the most straightforward solution for a misplaced power connector. Right?

If they shuffled pins around, the most likely target was simpler tracing on the PCB. Best case scenario (but unlikely), manufacturers can get rid of one PCB layer.

I have no idea, it was just a guess, I don't know PCB stuff well

 

[stimpy88]

There has to be more to this than a pin layout change. I openly love/hate nGreedia, and know they are constantly up to no good, and are always trying to hide it. I guess the early adopters will be the ones that decide whether its bad or not.

My money is on supporting DisplayPort 2.0+

 

[TheoneandonlyMrK]

Sales are poor. NV and AMD have miscalculated consumer enthusiasm for $1+k goods. You never know; leak this as a face saving exercise in humility, allows a price drop without admitting they got it wrong?

Though a one dollar decrease in BOM. How much can that equate to, really?

I can't see them passing that quid on to customers.

Odd all in.

 

[bug]

I can't see them passing that quid on to customers.

Odd all in.

Not odd at all. Silicon always goes through several revisions before it hits retail. These is probably some optimizations landing late and probably minor enough they weren't worth postposing release.

As for savings passed on, I'm not sure savings would save (pun) the current generation. I mean, I know I'm more stingy than most, but I'm not sure I'd touch the current generation at half the current MSRPs.

 

[TheoneandonlyMrK]

Not odd at all. Silicon always goes through several revisions before it hits retail. These is probably some optimizations landing late and probably minor enough they weren't worth postposing release.

As for savings passed on, I'm not sure savings would save (pun) the current generation. I mean, I know I'm more stingy than most, but I'm not sure I'd touch the current generation at half the current MSRPs.

There's virtually no saving.

All prior PCB designs need re doing to even fit the GPU due to pin out differences.

And this benefits Nvidia, because it sure doesn't benefit AIB's who invested in those designs on a margin that's been argued to be slim and now have to redo those designs.

And no performance difference, coming out of nvidia's PR department I'll believe that when it's proven externally Ty, they have prior in that regard.

 

[bug]

There's virtually no saving.

Not going by what we know so far. The new design may be simpler, easier to test or something like that, but nothing transpires from the official statement.

All prior PCB designs need re doing to even fit the GPU due to pin out differences.

And this benefits Nvidia, because it sure doesn't benefit AIB's who invested in those designs on a margin that's been argued to be slim and now have to redo those designs.

And no performance difference, coming out of nvidia's PR department I'll believe that when it's proven externally Ty, they have prior in that regard.

This isn't like they're changing PCBs individually, by hand. It's a small change in the template, validation and done. At the end of the day it's a BOM that's $1 less cheaper (subtract the changes to the template). That's for AIBs to pocket, because I don't see Nvidia charging $1 more for the new chips because the PCB got $1 cheaper.

 

[AdmiralThrawn]

NVIDIA making moves like this is killing small computer shops. If they drop the prices after all the computer stores have finally gotten stock we are going to lose thousands of dollars. AMD did the same thing with 7000 series. They release a product and computer shops like the one I work at order up enough stock for the holidays. Then 10 days later AMD price cuts the products after we have already payed for them so now we have to lose all of our profit because of a greedy company that wants to have their cake and eat it to. The sad part is if they do a price drop they will not lose any money, the vendors and stores will lose all of the difference.

 

[Dirt Chip]

1$ less- now that's a proper price cut. At last!