Wednesday, May 31, 2023
HomeNetworkingNvidia’s new Grace Hopper superchip to gasoline its DGX GH200 AI supercomputer

Nvidia’s new Grace Hopper superchip to gasoline its DGX GH200 AI supercomputer


Nvidia has unveiled a brand new DGX GH200 AI supercomputer, underpinned by its new Grace Hopper superchip and focused towards growing and supporting giant language fashions.

“DGX GH200 AI supercomputers combine Nvidia’s most superior accelerated computing and networking applied sciences to develop the frontier of AI,” Nvidia CEO Jensen Huang mentioned in a weblog put up.

The supercomputer, in response to Huang, combines the corporate’s GH200 Grace Hopper superchip and Nvidia’s NVLink and Swap System, to permit the event of huge language fashions for generative AI language purposes, recommender methods, and knowledge analytics workloads.

Nvidia’s DGX GH200 makes use of the NVLink interconnect know-how to mix 256 Grace Hopper superchips right into a single graphics processing unit (GPU) in an effort to extract “1 exaflop of efficiency and 144 terabytes of shared reminiscence — practically 500 occasions extra reminiscence than the earlier era NVIDIA DGX A100, which was launched in 2020.”

The chip maker will emulate the technique it took with its DGX Pods in making the brand new supercomputer accessible.

Earlier in March, Huang mentioned the corporate struck a deal to make its DGX methods accessible by way of a number of cloud suppliers, reasonably than putting in the mandatory {hardware} on-premises.

Presently, Microsoft, Meta, and Google Cloud have entry to the brand new supercomputer, the corporate mentioned.

Nvidia additionally mentioned the brand new Grace Hopper superchip that fuels the DGX GH200 AI supercomputer is in full manufacturing mode and methods with the superchip are anticipated to be accessible later this yr.

The corporate additionally mentioned that it was utilizing the brand new Grace Hopper superchip to assist SoftBank design next-generation distributed knowledge facilities that will likely be able to dealing with generative AI and 6G purposes.

These knowledge facilities will likely be distributed throughout Japan, the businesses mentioned in a weblog put up.

Earlier in March, the corporate had launched new knowledge processing models (DPUs) and GPUs, together with the BlueField 3 DPU.

Copyright © 2023 IDG Communications, Inc.

RELATED ARTICLES

LEAVE A REPLY

Please enter your comment!
Please enter your name here

- Advertisment -
Google search engine

Most Popular

Recent Comments