Abstract

The NVIDIA Enterprise Reference Architecture (Enterprise RA) with NVIDIA GB300 NVL72 and NVIDIA Spectrum-X Networking powers enterprise data centers for AI training and inference at massive scale, enabling real-time applications and trillion-parameter models.

This Enterprise RA is designed to solve the most challenging computational problems of today. Certified configurations based on this architecture can be deployed in enterprise data centers around the world.

This Enterprise RA uses the following NVIDIA technologies:

• NVIDIA GB300 NVL72

• NVIDIA BlueField®-3 DPUs

• NVIDIA ConnectX®-8 SuperNICs

• NVIDIA Spectrum™-X Ethernet Platforms

• NVIDIA AI Enterprise Software

• NVIDIA Mission Control

• NVIDIA NVLink and NVLink Switch

This document discusses the components that define the scalable and modular architecture of the Enterprise RA, which is built upon the building blocks of scalable units (SU), each containing 18 compute nodes with 72 GPUs. This Enterprise RA includes details regarding the SU design and specifics of Ethernet fabric topologies, NVLink networking, and storage system specifications.

The Enterprise RA is managed by NVIDIA solutions such as NVIDIA Mission Control, NVIDIA AI Enterprise, NVIDIA® CUDA®, and NVIDIA Magnum IO™. These technologies help keep the system running at the highest levels of availability and performance.