Skip to content

Comment on Europe enters the exascale supercomputing league with Jupiter

Comments

How do the gpu training clusters rank against the traditional “largest supercomputer” metric? Or is that comparing apples to oranges?

I think by some measures XAi already has the largest (200k H100 equivalent) and already constructing the 1 million gpu cluster.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.