PRODUCT

Organizations leading the future with endless challenges and changes, organizations delivering fast and convenient

AI/HPC Computing

 

DGX/HGXGPU Server

The key element of AI infrastructure, a global leader in AI computing.
Providing customized computing power for various AI demands, ranging from small-scale GPUs for AI research and development to the realm of hyperscale data centers

Leaping to world-class performance data center GPU

Deliver unprecedented performance and scalability for any workload.
The combination of multiple technological innovations provides more advanced computing resources than previous generations.

Data Sheetdownload
세계 최고 성능으로의 도약, 데이터센터 GPU1
세계 최고 성능으로의 도약, 데이터센터 GPU2
세계 최고 성능으로의 도약, 데이터센터 GPU3

The beginning of AI data center, NVIDIA GPU system

Data Center & Cloud

  • DGX/HGX Platform
  • NVIDIA DGX POD™

DGX H100

Flexible scaling and workload acceleration from small clusters to AI data centers

엑사스케일 고성능 컴퓨팅

NVIDIA Hopper Architecture

  • 8x H100 Tensor Core GPUs
  • 900GB/s 4th generation NVLink
  • Provides a total of 6,400 transistors

Optimized software stack

Integrated DGX software stack, including an AI-tuned base operating system, all necessary system software, GPU-accelerated applications, pre-trained models, and more

Optimized multi-user deployment

  • Split into 7 GPU instances with addition of Multi-Instance GPU (MIG) feature
  • GPU instances can be assigned to up to 56 users with 8 H100s

Built-in security

The most powerful security features for AI deployment through a multi-layered approach including baseboard management controller (BMC), CPU board, GPU board, self-encrypting drive, and secure boot.

DGX H100 Architecture

DGX H100 Architecture
1
close

8x NVIDIA H100 GPU
(total 640GB GPU memory)

18x NVIDIA® NVLink® connections per GPU, 900GB/s bi-directional GPU-GPU bandwidth

2
close

4x NVIDIA NVSwitches™

7.2TB/s bi-directional GPU-GPU bandwidth, 1.5x more than previous generation

3
close

10x NVIDIA ConnectX-7®400 Gb/s network interfaces

Maximum two-way network bandwidth

4
close

Dual Intel Xeon Platinum 8480C processors, 112 total cores, 2TB system memory

Powerful CPU for the most intensive AI tasks

5
close

30TB NVMe SSD

High-speed storage for maximum performance

Turnkey data center solutions
for the AI enterprise

By proactively optimizing and integrating compute, network, fabric, storage, and software,
we realize the most ideal stages of data center design, construction, and operation in an AI/HPC environment.

Performance

Powering the world's most powerful supercomputers

17 of the world's most powerful supercomputers in the TOP500 and Green500 are built on the NVIDIA DGX platform. These companies used DGX to scale their large-scale supercomputing infrastructure in just a few weeks.

World record breaking

NVIDIA DGX systems continue to set AI performance records in MLPerf, a set of benchmarks that measure machine learning training performance across a variety of use cases.

Setting a new standard for AI infrastructure

Organizations have used NVIDIA DGX SuperPOD™ to build the world's largest and most powerful AI supercomputers. This includes TOP500 and Green500 supercomputers that NVIDIA uses for its own research.

NVIDIA DGX SuperPOD

DGX SuperPOD
White Glove
Services

NVIDIA AI Enterprise
NVIDIA Base Command Platform
NVIDIA Base Command
NVIDIA Networking
DGX H100DGX A100
NVIDIA Base Command

Wide-ranging AI/HPC industry sectors utilizing GPU

Utilizes GPU-based parallel processing to execute large-scale application programs more efficiently, reliably, and quickly.

  • Supercomputing

    National supercomputer center, research institute, corporate IDC

    Supernova Explosion Research
    Earth interior mapping
    hurricane forecast
    simulation

  • Life Sciences / Precision Medicine

    National medical institutions / universities and large hospitals

    Video and image analysis/diagnosis
    Patient data analysis and risk measurement
    Medical consultation through chatbot
    New drug development
    Genetic mutation discovery

  • Internet Service provider

    Portal game company & cloud service company

    Image / Video Search / Classification
    voice recognition
    natural language processing
    Article Recommendations
    language translation
    Big data analysis

  • Communication / Media Shopping

    3 Telco companies and large shopping / media

    Product Recommendation
    image search
    natural language processing
    Real-time translation
    call center

  • Public Sector

    National government agencies and public institutions

    cyber security
    disaster response
    Intelligent Crime Prevention
    public service chatbot
    school research project

  • Manufacturing

    electronics & manufacturing companies and automobile companies

    autonomous driving
    smart factory
    Defect analysis through video analysis
    Big data analysis

  • Finance

    Domestic / overseas major banks and insurance / securities companies

    Prevent fraudulent transactions
    Robo-advisor (recommendations, trading, etc.)
    Chatbot / Call Center
    Big data analysis

  • Energy

    Major domestic electric power and oil refining institutions

    Development of thermal power generation and transmission and distribution technology
    High-precision computerized analysis simulation
    Development of new and renewable energy materials
    Robot design for power facilities
arrow_upward
close