Nvidia reveals Blackwell B200 GPU, the 'most powerful chip' for AI

It's terrifying to think about how many watts of power are being spent on generating useless trash

vasco · 2 years ago

Most human consumption is useless trash by some metric.

epistasis · 2 years ago

This is a bit vague, are you talking about this new GPU/LLMs, social media, industrial society, or something else?

reaperman · 2 years ago

Fun fact: each human uses an average 100W of power and generates mostly useless crap.

visarga · 2 years ago

> each human uses an average 100W of power

Does that include heating, electricity and fuel spent for the benefit of the human? How about infrastructure? Externalities are huge.

The article makes no mention at all of how much VRAM they might have. Other articles state plans for up to 192GB of HMB3e, and a power draw of 1000 watts.

Deleted Comment

sp332 · 2 years ago

I know it’s only FP4, but 1.4 exaflops in one rack is still crazy.

alecco · 2 years ago

With scaling and some tricks, FP4 inference can be very close to 16 bit.

And most software out there is still missing on FP8 lol

cyanydeez · 2 years ago

Doesn't that suggest the training method is flawed?

> next-gen NVLink switch that lets 576 GPUs talk to each other, with 1.8 terabytes per second of bidirectional bandwidth.

Well that's a pretty big step up from the 18 for the H100...

shouche · 2 years ago

Explained by AI

https://xplained.vercel.app/ask?q=Nvidia%20Blackwell

Additional details on vRAM and spec: https://xplained.vercel.app/ask?q=Nvidia%20Blackwell%20vRAM%...

Dead Comment

klysm · 2 years ago