NVIDIA has announced that its most powerful hardware for AI supercomputing is now the core of nine new supercomputers worldwide.

NVIDIA Grace Hopper KV

Together, these GH200 systems deliver 200 exaFLOPs of raw computing power while maintaining energy efficiency, enabling significant advancements in fields such as medicine and climate monitoring.

Here are some of the supercomputers, their locations, and their providers:

  • EXA1-HE – France – Hewlett Packard Enterprise
  • Alps – Switzerland – Hewlett Packard Enterprise
  • JUPITER – Germany
  • Delta AI – US
  • Miyabi – Japan
  • Isambard-AI – UK
  • Isambard 3 – UK
  • Unspecified systems at Los Alamos National Laboratory and Texas Advanced Computing Center – US

EXA1-HE, delivered in April, features Eviden’s patented BullSquana XH3000 warm-water cooling system, which cools 477 compute nodes. On the other hand, the Isambard-AI is currently in phase 1, housing an HPE Cray EX2500 supercomputer with 168 GH200 Superchips, with the remaining 5,280 chips arriving this summer to boost performance by approximately 32 times.

The full potential of these systems is realized through NVIDIA Quantum-2 InfiniBand networking, which links nodes into a giant logical system, and the AI and HPC software developed in-house.


Related Posts

Subscribe via Email

Enter your email address to subscribe to Tech-Critter and receive notifications of new posts by email.

Leave a Reply