HGX Reference Architecture

Powered by NVIDIA Tesla GPUs and NVLINK.

Standardized GPU Acceleration for AI Cloud Computing

Powered by NVIDIA® Tesla® GPUs and NVIDIA NVLink™, the HGX reference architecture standardizes the design of data centers accelerating AI in the cloud—from autonomous driving and personalized healthcare to superhuman voice recognition.

This architecture features a modular design that works seamlessly in existing data centers around the world, offering hyperscale facilities a quick, simple path to AI.

It taps into the power of eight Tesla GPUs interconnected with the NVLink hybrid cube to offer revolutionary performance with unprecedented configurability. Also, HGX is architected for NVIDIA Tesla® P100s and V100s.

Standardized GPU Acceleration for AI Cloud Computing

HGX in the industry’s leading AI servers

NVIDA® DGX-1™

MICROSOFT PROJECT OLYMPUS

FACEBOOK BIG BASIN

NVIDIA HGX Partner Program

NVIDIA delivers valuable resources that enable the world’s leading manufacturers –Foxconn, Huawei, Inspur, Inventec, Lenovo, Quanta, and Wistron–to more rapidly advance AI cloud computing. These include early access to the NVIDIA HGX reference architecture, NVIDIA GPU computing technologies, design guidelines, and more.

FOXCONN
Inventec
Quanta Computer
Wistron
Lenovo
Huawei
Inspur

Sign up for Data Center News

Learn about the latest in high performance computing, deep learning, and artificial intelligence.