Nvidia may divert Ada Lovelace desktop GPUs to notebooks to meet demand

[midian182]

Rumor mill: A new report claims that Nvidia is getting ready to divert some of its desktop GPU supply to notebooks as a way of keeping up with massive demand for the latter in China. The concern is that it could lead to another shortage of desktop graphics cards, but such a scenario seems very unlikely.

HardwareTimes writes that this is just a rumor from China, so take it with a pinch of salt. It's claimed that Nvidia is allocating some of its Ada Lovelace desktop chip capacity to notebooks, rather than increasing laptop GPU production to meet demand.

The source claims the RTX 4060 and RTX 4070 desktop cards will be impacted. We heard in April that Nvidia could cut the supply of RTX 4070 GPUs in response to slow sales, though we never saw much impact on the market if it did carry out these plans.

Both the RTX 4060 and its Ti versions have also failed to set the world on fire, especially in Japan. It suggests there is plenty of supply of the expensive cards, which some retailers have been offering at discounted prices as part of deals. In the case of the RTX 4060 Ti, it only just entered the Steam chart, and with a comparatively tiny user share.

The top-performing graphics cards among Steam survey participants last month

As PC Gamer notes, however, a tighter supply of these cards could enable their retail prices to stay at their standard high level, with no deals or discounts on offer.

Any talk of supply restrictions brings back bad memories of the graphics card shortage that saw prices reach three times their MSRP at one point. The pandemic, chip issues, and high cryptocurrency prices combined in a perfect storm that made purchasing a graphics card a matter of luck and money – lots of money.

Even if the RTX 4060 and 4070 did see their supply reduced, the limited demand for these expensive cards means we might not even notice.

There's also AMD's upcoming Radeon RX 7800 and RX 7700 to consider. They're rumored to be officially announced at Gamescom 2023, which starts next week, and are expected to compete with or even outperform Nvidia's midrange products while sporting a cheaper MSRP. That could prompt Team Green to introduce some price cuts to the RTX 4060/Ti and 4070.

Permalink to story.

https://www.techspot.com/news/99794-nvidia-may-divert-lovelace-desktop-gpu-supply-notebooks.html

 

[Haaky]

Yea they have to ship there stuff off to China cause nobody wants to buy there over priced garbage here
What do you expect from a scumm bag , money grubbing company

 

[DSirius]

Funny and subtle subtitle question.

 

[manowar51]

Nvidia are using the 4nm process for their RTX 4000 CPUs which is quite expensive and probably a factor as to why the prices are so high. They probably worked out that they would get more money per wafer if they use this node for laptop GPUs. I cant see them going much lower on the desktop RTX 4000 series. They will wait for the 5000 series and be like "look its 10% cheaper and 10% faster" and I'm sure the market will eat that bs up.

On the other hand Radeon is still using the cheaper 5nm & 6nm process nodes for their GPUs so they should have more wiggle room on pricing. They should be able to lower prices even further if things get really bad. I think its quite impressive that Radeon competes with Geforce despite the larger die sizes.

P.S. I love that the 6th most popular GPU on steam is a Sandy Bridge iGPU. Thats a relic from the golden age of PC gaming, where a new console cost $300 and gave you 20 fps at 600p but you could build a budget gaming pc for $500 that would run games at 1080p60. Now $500 is just your motherboard!

 

[sheady]

MLID talked about Nvidia reducing lovelace desktop supply almost 2 weeks ago

 

[i like foxes]

Haaky , why do you think they re overpriced ? Here , a lot of foods have doubled their price but the GPUs not . Not that I wish it happened .

 

[takaozo]

Nvidia loves China market so much that they cant spare any ocasion to wine about the loss the trade war is impacting them. I wouldnt be surprised if Nvidia will make another company somewhere like Belize/Cayman/Cyprus and resell to China from that third party company.

 

[SalmanIsHere]

Nvidia are using the 4nm process for their RTX 4000 CPUs which is quite expensive and probably a factor as to why the prices are so high. They probably worked out that they would get more money per wafer if they use this node for laptop GPUs. I cant see them going much lower on the desktop RTX 4000 series. They will wait for the 5000 series and be like "look its 10% cheaper and 10% faster" and I'm sure the market will eat that bs up.

- Nvidia never released any gpu that is 10% faster vs their old counterpart. Also, they are using same 5nm like Amd. For marketting purpose Jensen calls 5nm as 4N. Just like he called Samsung 10nm as 8N. N means nvidia.

On the other hand Radeon is still using the cheaper 5nm & 6nm process nodes for their GPUs so they should have more wiggle room on pricing. They should be able to lower prices even further if things get really bad. I think its quite impressive that Radeon competes with Geforce despite the larger die sizes.

- Both using same 5nm. Amd gpus actually costs much more to make. For example- AD103 cut 4080 is 380mm, 256 bit 16gb vs Full N31 520mm 384 bit 24 gb 7900xtx. 7900xtx matches in Raster but is 30% slower in Max settings(Rt+Raster).

 

[yRaz]

- Nvidia never released any gpu that is 10% faster vs their old counterpart. Also, they are using same 5nm like Amd. For marketting purpose Jensen calls 5nm as 4N. Just like he called Samsung 10nm as 8N. N means nvidia.

On the other hand Radeon is still using the cheaper 5nm & 6nm process nodes for their GPUs so they should have more wiggle room on pricing. They should be able to lower prices even further if things get really bad. I think its quite impressive that Radeon competes with Geforce despite the larger die sizes.

- Both using same 5nm. Amd gpus actually costs much more to make. For example- AD103 cut 4080 is 380mm, 256 bit 16gb vs Full N31 520mm 384 bit 24 gb 7900xtx. 7900xtx matches in Raster but is 30% slower in Max settings(Rt+Raster).

Do you have a source for absolutely any part of what you said?

 

[toooooot]

Sure, go for it. I am still not paying for sub class cards costing the same.

 

[Mr Majestyk]

Nvidia loves China market so much that they cant spare any ocasion to wine about the loss the trade war is impacting them. I wouldnt be surprised if Nvidia will make another company somewhere like Belize/Cayman/Cyprus and resell to China from that third party company.

Well the CEO is Chinese and probably besties with Xi. Huang was once a humble engineer, now's he's like all money grubbing scumbag billionaires.

 

[SalmanIsHere]

Do you have a source for absolutely any part of what you said?

It is google search away.

*** 4n is actually 5nm - https://www.techgoing.com/nvidia-cl...ong Kong media,number of media writing errors.

*** 7900xtx spec -https://www.techpowerup.com/gpu-specs/radeon-rx-7900-xtx.c3941

*** 4080 spec- https://www.techpowerup.com/gpu-specs/geforce-rtx-4080.c3888

*** 4080 is 30%+ faster in Max Settings vss 7900xtx (Rt+Raster On) - https://www.computerbase.de/2023-01...chmarks_mit_und_ohne_raytracing_in_3840__2160

 

[neeyik]

*** 4n is actually 5nm - https://www.techgoing.com/nvidia-cl...ong Kong media,number of media writing errors.

The 4N thing is really a typo by Nvidia, or at the very least, an incorrect use of TSMC's terminology. Nodes used specifically for Nvidia's products have always followed the standard TSMC nomenclature, followed by an addendum to indicate its exclusivity -- e.g. Turing was made on 16FFN, with the standard variant being 16FF.

Its '5nm' family of process nodes are variants of the same methodology, with variations in the front-end, middle, and back-end of the production, as well as different levels of EUV usage. So N5, N5P, N5HPC, N4, N4P, and N4X are all similar (each one is IP compatible with the other) but they're not identical -- N4 has a slightly higher density than N5 (roughly 5%) and lower/higher power/clock scaling.

Nvidia is most likely to be using the N4P variant, to achieve the power and clock goals for its products, as both aspects are better than those with N5. It's also worth noting that N5 has been in volume manufacturing far longer than N4 has, so not only is the process used to fabricate the GCD for the Navi 31 GPU different from that used by Nvidia, but it's also cheaper.