NVIDIA DGX™ A100

First AI system built on new NVIDIA A100 GPU
  • 8x NVIDIA A100 Tensor Core GPUs, which deliver unmatched acceleration
  • Tensor Float 32 (TF32) for 20x higher FLOPS
  • 8x NVIDIA A100 80 GB GPUs
  • Available with up to 640 gigabytes (GB) of total GPU memory
  • 8x SinglePort Mellanox ConnectX-6 VPI 200Gb/s HDR InfiniBand
Request

THE CHALLENGE OF SCALING ENTERPRISE AI

Every business needs to transform itself using artificial intelligence (AI), not just to survive, but to thrive in tough times. However, the enterprise needs an AI infrastructure platform that improves on traditional approaches that in the past included slow computing architectures separated by analytics, training and inference workloads. The old approach introduced complexity, drove up costs, limited scalability and was not ready for modern AI. Enterprises, developers, data scientists, and researchers need a new platform that unifies all AI workloads, simplifies infrastructure, and accelerates ROI.

THE UNIVERSAL SYSTEM FOR ALL KI WORKLOADS

NVIDIA DGX™ A100 is the universal system for all AI workloads - from analytics to training to inference. The DGX A100 sets new standards for compute density, packing 5 petaFLOPS of AI performance into a 6U form factor and replacing legacy compute infrastructure with a single, unified system. The DGX A100 also provides the unprecedented ability to allocate compute power in a fine-grained manner by leveraging the Multi-Instance GPU (MIG) feature of the NVIDIA A100 Tensor Core GPU, which enables administrators to allocate resources that are right-sized for specific workloads.
The DGX A100 is available with up to 640 gigabytes (GB) of total GPU memory, which increases performance by up to three times for large training jobs and doubles the size of MIG instances. This allows the DGX A100 to handle the largest and most complex jobs as well as the simplest and smallest. The DGX A100 runs the DGX software stack and optimized software from NGC. The combination of dense compute performance and complete workload flexibility makes the DGX A100 an ideal choice for both single-node deployments and large-scale Slurm and Kubernetes clusters deployed with NVIDIA DeepOps.

IT DEPLOYMENT SERVICES AVAILABLE

Want to shorten time to insights and accelerate ROI from AI? Let our professional IT team accelerate, deploy and integrate the world's first 5 petaFLOPS AI system, NVIDIA® DGX™ A100, seamlessly and non-disruptively into your infrastructure with 24/7 support.
Get the results and outcomes you need:
  • Site analysis, readiness, pre-testing and staging
  • Access to dedicated engineers, solution architects and support technicians
  • Deployment planning, scheduling and project management
  • Shipping, logistics management and inventory provisioning
  • On-site installation, on-site or remote configuration of software
  • Post-deployment check-up, support, ticketing, and maintenance
  • Lifecycle management including design, upgrades, recovery, repair and disposal
  • Rack and stack and integration services and multi-site deployment
  • Customized break-fix and managed service contracts

NVIDIA A100 Tensor Core GPU

  • 8X NVIDIA A100 GPUS WITH UP TO 640 GB TOTAL GPU MEMORY
    12 NVLinks/GPU, 600 GB/s GPU-to-GPU bi-directonal bandwidth
  • 6X NVIDIA NVSWITCHES
    4.8 TB/s Bi-directional Bandwidth, 2X More than Previous Generation NVSwitch
  • 10X MELLANOX CONNECTX-6 200 Gb/s NETWORK INTERFACE
    500 GB/s Peak Bi-directional Bandwidth
  • DUAL 64-CORE AMD CPUs AND 2 TB SYSTEM MEMORY
    3.2X More Cores to Power the Most Intensive AI Jobs
  • 30 TB GEN4 NVME SSD
    50 GB/s Peak Bandwidth, 2X Faster than Gen3 NVME SSDs

Technology Inside NVIDIA DGX A100

multi-instance GPU (MIG)
The eight A100 GPUs in the DGX A100 can be configured into up to 56 GPU instances, each fully isolated with its own memory, cache and high-bandwidth processing cores.
nVLink& NVSwitch
Third-generation NVIDIA® NVLink™ doubles direct GPU-to-GPU bandwidth to 600 GB/s, which is nearly 10x faster than PCIe Gen4 and next-generation NVIDIA NVSwitch™, which is 2x faster than the previous generation.
infiniband
The new Mellanox ConnectX-6 VPI HDR InfiniBand/Ethernet adapters run at 200 gigabits per second (Gb/s) to create a high-speed fabric for large AI workloads.
optimized software stack
Integrated DGX software stack, including a base operating system tuned for AI, all required system software and GPU-accelerated applications, pre-trained models, and more.
Integrated security
The most robust security for AI deployments with a layered approach spanning the Baseboard Management Controller (BMC), CPU board, GPU board, self-encrypted drives, and secure boot.

ESSENTIAL BUILDING BLOCK OF THE KI DATA CENTER

The universal system for any AI workload

A universal building block to run any workload at any time -
from analytics to AI training to inferencing.

Integrated access to AI expertise

Accelerate AI transformation with NVIDIA DGXperts to maximize the value of your DGX investment.

Game changing performance for innovators

Offers unprecedented acceleration with eight A100 GPUs and is fully optimized for NVIDIA CUDA-X™ software and the end-to-end NVIDIA data center solution stack.

Unmatched data center scalability

Build a world-class AI infrastructure that is scalable to keep pace with demand.

GPUs

NVIDIA DGX A100 640GB

  • 8x NVIDIA A100 80 GB GPUs
GPU Memory

640 total

CPU
Dual AMD Rome 7742, 128 cores total, 2.25 GHz (base), 3.4 GHz (max boost)
NVIDIA NVSwitches
6
System Power Usage
6.5kW max
System Memory

2 TB

Networking

8x single port Mellanox ConnectX-6 VPI
200Gb/s HDR InfiniBand
2x dual-port Mellanox ConnectX-6 VPI
10/25/50/100/200 Gb/s Ethernet

Storage

OS: 2x 1.92 TB M.2 NVME drives
Internal Storage: 30 TB (8x 3.84 TB) U.2 NVMe drives

Software

Ubuntu Linux OSAlso
supports: Red Hat Enterprise Linux CentOS

System Weight

271.5 lbs (123.16 kgs) max

System Dimensions
  • Height: 10.4 in (264.0 mm)
  • Width: 19.0 in (482.3 mm) max
  • Length: 35.3 in (897.1 mm) max
Packing System Weight
359.7 lbs (163.16 kgs) max