In the event you’ve ever been left annoyed by the mere 24GB of VRAM in an RTX 4090D or discovered the pathetic 16GB hooked up to an RTX 4080 Tremendous to be completely under you, there may very well be an answer accessible, in case you’re resourceful. That’s as a result of one Chinese language Cloud AI firm has been proven utilizing these GPUs however with double their regular VRAM allocation. So, in case you ask actually, actually properly, they may kind you out with a brand new double VRAM Nvidia graphics card too.
Nvidia isn’t set to launch any new finest graphics card contenders till the a lot anticipated RTX 5000 collection launch on the tail finish of this 12 months or begin of subsequent 12 months. Nevertheless, it appears that evidently hasn’t stopped this unnamed firm from discovering its personal strategy to eke out much more efficiency from its Nvidia playing cards.
The leaked data comes through X person @bdsqlsz who posted on the service a number of screenshots of methods operating what declare to be RTX 4090D and RTX 4080 Tremendous GPUs with double their regular quantity of VRAM. Which means the RTX 4090D is sporting 48GB whereas the RTX 4080 packs in 32GB.
The playing cards are being utilized by a Chinese language Cloud computing service the place the additional reminiscence is beneficial for AI computing functions, reasonably than it getting used to retailer gaming information.
Notably, the PCB that homes the RTX 4090D GPU isn’t usually one which’s capable of accommodate sufficient reminiscence chips to have the ability to supply 48GB of VRAM, because it solely homes the chips on one facet of the board. As an alternative, the corporate is more likely to have used outdated RTX 3090 Ti PCBs which have reminiscence places on either side of the board, with these boards additionally recognized to be appropriate with RTX 4090 GPUs. Nevertheless, with no footage of the playing cards or different data to go on, we are able to’t know for positive.
You might be questioning why an organization like this is able to go to such efforts to improve an RTX 4090D (or certainly the RTX 4080 Tremendous) when the RTX 4090 is extra highly effective and there can be found much more highly effective devoted server GPUs that may inherently be geared up with extra VRAM.
Answering these in reverse order, utilizing consumer-grade chips just like the RTX 4090D and RTX 4080 Tremendous is simply less expensive than shopping for the server-grade choices. In the meantime, the RTX 4090 isn’t an choice to be used in China in the mean time on account of US export restrictions. Certainly, that’s why the RTX 4090D exists – it’s a cut-down variant of the 4090 that’s authorized by the US authorities to be used in China. This has led to many Chinese language corporations having to turn out to be further resourceful to seek out the highly effective playing cards they want.
Realistically, your probabilities of getting maintain of a card like this are very slim, because it’s pretty unlikely such an organization would be prepared to promote you one, and getting the product shipped from China is likely to be tough too. You could possibly endeavor to do the mod your self or make use of another modder to tackle the duty, but when your purpose is to enhance gaming efficiency, it’s an costly, dangerous choice we wouldn’t suggest. You’ll get principally no efficiency enchancment in video games, as virtually no sport max out the 24GB of reminiscence on an RTX 4090D as it’s.
As an alternative, you’re a lot better off simply splashing the money on a full-fat RTX 4090, for a begin. Or, in case you’re truly on the lookout for a card that provides respectable worth whereas additionally offering blazing efficiency, the RTX 4080 Tremendous is the cardboard to go for.
