Let's talk

NVIDIA DGX H200

NVIDIA Unveils DGX H200: A groundbreaking AI supercomputing platform with 4x the bandwidth of previous models.

NVIDIA DGX H200 powers the future of AI with Next-Gen H200 GPUs and unified GPU memory architecture.

2x Intel Xeon 8480C (56 cores, 2.0GHz), 8x NVIDIA H200 SXM5 141GB, 2TB RAM, 30TB NVMe (Data), 8x 400Gb HDR Infiniband, 4x 200Gb Ethernet, 3 years support

 
DGX_A100_2
DGX_A100_3-600x500
NVIDIA_DGX_H100_7u-600x500
NVIDIA_DGX_H100_8u
NVIDIA_DGX_H100_9u-600x500
NVIDIA_DGX_H100_10u-600x500
NVIDIA_DGX_H100_11u-600x500

// solution overview

The Universal System for Every AI Workload

As the fourth generation of the world’s first purpose-built AI infrastructure, DGX H200 is designed to be the centerpiece of an enterprise AI center of excellence.

It’s a fully optimized hardware and software platform that includes full support for the new range of NVIDIA AI software solutions, a rich ecosystem of third-party support, and access to expert advice from NVIDIA professional services.

 

 

performance
Configuration flexibility
TCO
premium support

// extreme performance

Computing on the most powerful NVIDIA GPUs

Up to 1,128 GB

total GPU memory

NVIDIA AI software solutions The Cornerstone of Your AI Center of Excellence

DGX H200 is a fully integrated hardware and software solution on which to build your AI Center of Excellence. It includes NVIDIA Base Command™ and the NVIDIA AI Enterprise software suite, plus expert advice from
NVIDIA DGXperts.

Leadership-Class Infrastructure on Your Terms

Get the power of DGX H200 in a multitude of ways that fit your business: on premises, co-located, rented from managed service providers, and more. And with DGX-Ready Lifecycle Management, organizations get a predictable financial model to keep their deployment at the leading edge.

Break Through the Barriers to AI at Scale

NVIDIA DGX H200 features 6X more performance, 2X faster networking, and high-speed scalability. Its architecture is supercharged for the largest workloads such as generative AI, natural language processing, and deep learning recommendation models.

NVIDIA H200 SXM5 GPUs

8x NVIDIA H200 with 1128GB total GPU memory, 18x NVLink connections per GPU with 900 GB/s GPU-to-GPU bandwidth

NVLink & NVSwitch

4x NVIDIA NVSwitches with 7.2TB/s of bidirectional bandwidth

30TB Gen4 NVMe SSD

8× 3.84 TB with 50GB/s of peak bandwidth, 2x faster than Gen3 NVMe SSDs and 2× 1.92 TB NVMe for OS

// Extreme Performance

NVIDIA DGX H200

Dual Intel Xeon Platinum 8480C CPU

A total of 112 processor cores and 2TB of system memory

High-speed fabric

10x NVIDIA ConnectX-7 with 400Gb/s InfiniBand / Ethernet network interface

Optimized Software Stack

DGX OS, all necessary system software, GPU-accelerated applications and pre-trained models

// Software & Services

Fastest Time to Solution

DGX H200 integrates a tested and optimized DGX software stack, including an AI-tuned base operating system, all necessary system software, and GPU-accelerated applications, pre-trained models, and more from NGC.

NVIDIA Enterprise Services provide support, education, and professional services for your DGX infrastructure. With NVIDIA experts available at every step of your AI journey, Enterprise Services can help you get your projects up and running quickly and successfully.

caffe
Jupyter
keras
Pytorch
tensorflow

// DGX software

NVIDIA DGX Systems Provide Full AI Software Stack

01.

NVIDIA DGX server

The universal system for all AI workloads – from analytics to training to inference. DGX H200 sets a new bar for compute density, packing 32 petaFLOPS of AI performance into a 8U form factor, replacing legacy compute infrastructure with a single, unified system.

02.

NVIDIA DGX OS

NVIDIA DGX Systems comes with pre-installed NVIDIA DGX OS software stack – Ubuntu or RedHat Enterprise Linux, optimized drivers (GPU, networking), Docker engine, container management, system management, datacenter management tools and security features.

03.

NVIDIA AI Enterprise

NVIDIA AI Enterprise offers end-to-end, secure, cloud-native suite of AI software which provides a large catalog of optimized and up-to-date data science libraries, tools, pre-trained models, frameworks, and more. The entire software stack is fully covered with NVIDIA enterprise support.

04.

NVIDIA Base Command

NVIDIA Base Command provides Kubernetes (platform for orchestration of application containers), Slurm, and Jupyter Notebook environments, delivering an easy-to-use, enterprise-proven scheduling and orchestration solution based on well-established enterprise standards.

05.

NGC applicatioin containers

NVIDIA NGC is public catalog of the most used AI frameworks, GPU accelerated applications, pre-trained AI models or workflows. You can run your AI workload right after delivery of your DGX system. NGC catalog is regurarly updated to always get the latest tuned software components for highest performance of your AI applications.
 NVIDIA DGX H200 1128 GB
GPUs8× NVIDIA H200 SXM5 141GB
GPU memory1128 GB
CPUDual Intel Xeon Platinum 8480C CPU, (112 cores) 2.00 GHz (Base), 3.80 GHz (Max Boost)
Performance32 PetaFLOPS (FP8)
# CUDA cores
135 168
# Tensor cores
4 224
Multi-instantce GPU56 instances
RAM2 TB
HDDOS: 2× 1.92 TB NVMe
DATA: 30 TB (8× 3.84 TB) NVMe
Networking8x single-port ConnectX-7 VPI 400 Gb/s InfiniBand/ 200Gb/s Ethernet
2x dual-port ConnectX-7 VPI 400 Gb/s InfiniBand/ 200Gb/s Ethernet
Max. power usage
~10,2 kW max
Form factorrack, 8U
Datasheet
Download

Resources

Continue Exploring

 

Technical walkthrough on how to get started with NVIDIA Quantum-X800.

Download PDF

NVIDIA Silicon Photonics Resources

Download PDF

NVIDIA DGX H200

Related Products