TechFlow news — NVIDIA today announced a new class of large-memory AI supercomputer, connecting 256 Grace Hopper Superchips into a single massive model with 1 exaflop of performance and 144TB of GPU memory, designed to power generative AI, recommender systems, and data analytics.
This computer features an enormous shared memory space, using NVLink interconnect technology and the NVLink Switch System to connect all superchips, delivering 1 exaflop of performance and 144 TB of shared memory—nearly 500 times the memory capacity of NVIDIA's previous-generation DGX A100 system introduced in 2020.
The DGX GH200 supercomputer integrates NVIDIA’s most advanced accelerated computing and networking technologies, enabling the development of next-generation giant models for generative AI language applications, recommender systems, and data analytics workloads.
