Hacker News Reader

Comment by ENGNR | Hacker News Reader

They have some pretty hefty minimum use and right to use options tied to the nuclear reactor though. And it says the intention is to consume 960MW, which a very quick search says is roughly equivalent to the entire output of a typical nuclear reactor.

View on HN

The NPP has 2x 1350 MWe boiling water reactors

960MWe is 71% of one of them.

Does anyone know how many NVIDIA data center GPUs can be run off 960MW?

https://www.anandtech.com/show/16792/nvidia-unveils-pcie-ver...

The 80GB version runs 300W on PCIe and 400W on SXM4. (Which is my third reason to hate the current wave of "AI" bullshit, the first two being industry scale copyright violation and barely better than slave labor).

Stop complaining about usage, from crypto to AI, they all have relevant uses to push the world forward.

Think about how to increase quality of life and output, the companies will automatically increase efficiency to increase profits and reduce costs.

> Stop complaining about usage, from crypto to AI, they all have relevant uses to push the world forward.

Can you elaborate on what you have in mind when you clam that crypto is "pushing the world forward"?

It pushes the world closer towards heat death of the Universe.

Since hubble said space is expanding, that means there will always be more room for entropy to increase. We will not have a heat death. Rather we are heading towards isolation and solitary confinement. I wonder if ET is behind this

Forward to what ?

> companies will automatically increase efficiency to increase profits

My sarcasm detector went off when I got to this bit.

Do you think there is always a positive relationship between production/consumption and quality of life? It seems to me this statement confuses what we should be ultimately optimizing for.

100,000 GPU based nodes = 100MW (Look at Frontier supercomputer usage)

So, 1 million GPU-based servers, which is just double of what Meta is planning to use for its own supercomputer. So, not that much.

As I hear, for ML machines is normal to have 16 GPUs, if each consume 250W, totally 4KW per machine, so need 250,000 machines for 1GW (for 40MW would be 10000 machines, so 240,000 for 960MW).

240k machines is a big number, I myself seen DC for 40k machines, but they talking about campus, so reasonable assume, there would be not one DC but few DC buildings.

I have googled, but at the moment have only this (will add more later):

50 MW

www.google.com/maps/search/Apple+Mesa/@33.3470187,-111.6092537,1311m/data=!3m1!1e3!5m1!1e4?entry=ttu

As I understand it, in fact 960MW is quite close to 1 jigawatt, which is the typical energy required to power a Delorean-class time machine.

Reply on news.ycombinator.com