THIS is why we can’t have nice things…
Nvidia delivers first Vera Rubin AI GPU samples to customers — 88-core Vera CPU paired with Rubin GPUs with 288 GB of HBM4 memory apiece
Submitted 1 day ago by RegularJoe@lemmy.world to technology@lemmy.world
Comments
zebidiah@lemmy.ca 7 hours ago
Mynameisallen@lemmy.zip 1 day ago
This is what all the parts we wanted went to
Earthman_Jim@lemmy.zip 1 day ago
Yeah, I wonder how long it will take them to clue in that no one wants to trade gaming for an AI fucking girlfriend ffs…
Mynameisallen@lemmy.zip 1 day ago
Until the money stops pouring in I suppose
setsubyou@lemmy.world 1 day ago
I mean if they came with a cool android body we could talk about it. It should at least be able to do cleaning and cooking. Otherwise my wife won’t like it.
roofuskit@lemmy.world 1 day ago
Don’t worry, you can rent them for $30 a month and stream all your video games.
Mynameisallen@lemmy.zip 20 hours ago
Not even, just the ones they deign to allow
gnawmon@ttrpg.network 1 day ago
so that’s why my 5070 laptop has 8 GBs of VRAM…
my old 1080 also had 8 GBs of VRAM
kittenzrulz123@lemmy.dbzer0.com 5 hours ago
Your 5070 laptop has 8gb of vram? My desktop 3060 has 12gb of vram and its not even the TI version.
gnawmon@ttrpg.network 4 hours ago
Cocodapuf@lemmy.world 20 hours ago
Jesus fucking Christ, 288GB. And this is why I can’t have 16?
Corkyskog@sh.itjust.works 6 hours ago
And you have to buy a rack of them with 72 of them.
xxce2AAb@feddit.dk 1 day ago
Goodbye, sweet hardware. You deserved better and so did we.
phoenixz@lemmy.ca 17 hours ago
And none of us will be allowed to have them
Only datacenters and only fortune 500 companies will be able to use anything Nvidia
eleitl@lemmy.zip 46 minutes ago
You can’t do much with them, unless you’re into deep leaning. And the power bill would bankrupt you. I wish I had a Cerebras box, but even the smallest one is 20 kW, liquid cooled.
Corkyskog@sh.itjust.works 6 hours ago
I mean if you have the 3 million to spend on a rack of them, I am sure they would allow you to have them.
I do wonder what happens a few years down the road when everyone are replacing their gpus with latest and greatest variants what happens to the old racks? Do they get sold for pennies on the dollar because everyone else doing AI wants cutting edge?
eleitl@lemmy.zip 45 minutes ago
The failure rate is high for ML GPUs. The hardware is effectively consumables.
RegularJoe@lemmy.world 1 day ago
Nvidia’s Vera Rubin platform is the company’s next-generation architecture for AI data centers that includes an 88-core Vera CPU, Rubin GPU with 288 GB HBM4 memory, Rubin CPX GPU with 128 GB of GDDR7, NVLink 6.0 switch ASIC for scale-up rack-scale connectivity, BlueField-4 DPU with integrated SSD to store key-value cache, Spectrum-6 Photonics Ethernet, and Quantum-CX9 1.6 Tb/s Photonics InfiniBand NICs, as well as Spectrum-X Photonics Ethernet and Quantum-CX9 Photonics InfiniBand switching silicon for scale-out connectivity.
TropicalDingdong@lemmy.world 1 day ago
288 GB HBM4 memory
jfc…
Looking at the spec’s… fucking hell these things probably cost over 100k.
I wonder if we’ll see a generational performance leap with LLM’s scaling to this much memory.
AliasAKA@lemmy.world 1 day ago
Current models are speculated at 700 billion parameters plus. At 32 bit precision (half float), that’s 2.8TB of RAM per model, or about 10 of these units. There are ways to lower it, but if you’re trying to run full precision (say for training) you’d use over 2x this, something like maybe 4x depending on how you store gradients and updates, and then running full precision I’d reckon at 32bit probably. Possible I suppose they train at 32bit but I’d be kind of surprised.
in_my_honest_opinion@piefed.social 1 day ago
Fundamentally no, linear progress requires exponential resources. The below article is about AGI but transformer based models will not benefit from just more grunt. We’re at the software stage of the problem now. But that doesn’t sign fat checks, so the big companies are incentivized to print money by developing more hardware.
https://timdettmers.com/2025/12/10/why-agi-will-not-happen/
Also the industry is running out of training data
https://arxiv.org/html/2602.21462v1
What we need are more efficient models, and better harnessing. Or a different approach, reinforced learning applied to RNNs that use transformers has been showing promise.
boonhet@sopuli.xyz 1 day ago
LLMs can already use way more I believe, they don’t really run them a single one of these things.
The HBM4 would likely be great for speed though.
Cocodapuf@lemmy.world 20 hours ago
panda_abyss@lemmy.ca 1 day ago
Yeah they’re going to cost as much as a house.
I think we’ll see much larger active portions of larger MOEs, and larger context windows, which would be useful.
The non LLM models I run would benefit a lot from this, but I don’t know of I’ll ever be able to justify the cost of how much they’ll be.
yogurtwrong@lemmy.world 20 hours ago
The buzzwords make my head hurt. Sounds like a copypasta
in_my_honest_opinion@piefed.social 14 hours ago
Almost like an LLM wrote it…
redsand@infosec.pub 17 hours ago
Brick them all 🧱
RizzRustbolt@lemmy.world 18 hours ago
But can it run Crysis?
elucubra@sopuli.xyz 16 hours ago
Can it run Doom?
Earthman_Jim@lemmy.zip 1 day ago
who. fucking. cares.
Hadriscus@jlai.lu 1 day ago
Can’t wait for it to hit secondhand market in november
Cocodapuf@lemmy.world 20 hours ago
So we can do what? De solder the individual ram chips and populate them on custom dimms?
Pass.
in_my_honest_opinion@piefed.social 14 hours ago
You scoff but this is already being done in China. They desolder good chips from bad cards and add them to a mule card.
vaultdweller013@sh.itjust.works 18 hours ago
Bringus is gonna make a weird gaming computer by shoving one into a movie rental kiosk.
fubarx@lemmy.world 1 day ago
Question is, how long before it makes it to the next DGX Spark? Some people don’t have $10B to burn.
LoremIpsumGenerator@lemmy.world 4 hours ago
So this is where our future ram buy went into? Fuck this planet then 🤣
eleitl@lemmy.zip 49 minutes ago
HBMx is a different product than DDRx/GDDRx, though parts of the fabbing are probably shared.