NVIDIA DGX POD™

NVIDIA DGX POD and DGX BasePOD

The Infrastructure Foundation for Scale Out AI

NVIDIA DGX Platform

NVIDIA DGX Deployments for Scale Out AI

NVIDIA DGX POD and DGX BasePOD solutions provide the underlying infrastructure (compute, networking, and storage) + software to accelerate the rapid deployment and execution of AI workloads. DGX POD deployments also include the AI data-plane/storage that can keep up with these workloads: with the capacity for large training datasets, expandability for growth, and the high storage throughput required for scale-out AI.

1-2 NVIDIA DGX A100s with DDN AI400X2

First Deployment of AI Compute & AI Ready-Parallel Storage

Specifications

8-16 NVIDIA H100 Tensor Core GPUs for AI throughput
NVIDIA NGC Containers and NVIDIA AI Enterprise for rapid deployment of AI workloads
NVIDIA Bright Cluster Manager for workload orchestration
DDN AI400X2 parallel storage appliance with throughput up to 90GB/sec and 3 million IOPS (various data capacities available)
Full parallel filesystem and Storage Management GUI
Total of 640GB or 1280GB of GPU memory
Seamless AI compute scaling: add DGX systems to increase AI performance (existing DDN AI400X2 has headroom to continue to scale data throughput)
Seamless storage throughput & capacity scaling: add DDN AI400X2 appliances to double data bandwidth or grow data capacity (DDN EXA Lustre filesystem is already built for expansion)
Superior Performance with GPUDirect Storage: realize a direct data path from storage to GPU over InfiniBand. Delivers faster performance for multiple users on a single DGX and on scale-out multi-DGX deployments

4 NVIDIA DGX H100s with Choice of Storage

Full Rack AI Infrastructure with Compute, AI Ready Storage, and Software

Specifications

32 NVIDIA H100 Tensor Core GPUs for maximum AI throughput
NVIDIA NGC Containers and NVIDIA AI Enterprise for rapid deployment of AI workloads
NVIDIA Base Command™ for workload orchestration
Choice of DDN AI400X2, Weka, VAST, or IBM Storage Scale parallel storage
Full parallel filesystem and Storage Management GUI
Total of 2560GB of GPU memory
Seamless AI compute scaling: double the size of your BasePOD to increase AI performance
Validated Configuration: Scale-out AI performance of design is validated by NVIDIA and each storage vendor

8 NVIDIA DGX H100s with Choice of Storage

Multi-Rack AI Infrastructure with Compute, AI Ready Storage, and Software

Specifications

64 NVIDIA H100 Tensor Core GPUs for maximum AI throughput
NVIDIA NGC Containers and NVIDIA AI Enterprise for rapid deployment of AI workloads
NVIDIA Base Command™ for workload orchestration
Choice of DDN AI400X2, Weka, VAST, or IBM Storage Scale parallel storage
Full parallel filesystem and Storage Management GUI
Total of 2560GB of GPU memory
Seamless AI compute scaling: double the size of your BasePOD to increase AI performance
Validated Configuration: Scale-out AI performance of design is validated by NVIDIA and each storage vendor

16 NVIDIA DGX H100s with Choice of Storage

Large Organization AI Infrastructure with Compute, AI Ready Storage, and Software

Specifications

128 NVIDIA H100 Tensor Core GPUs for maximum AI throughput
NVIDIA NGC Containers and NVIDIA AI Enterprise for rapid deployment of AI workloads
NVIDIA Base Command™ for workload orchestration
Choice of DDN AI400X2 (2 units), Weka, VAST, or IBM Storage Scale parallel storage
Full parallel filesystem and Storage Management GUI
Total of 5120GB of GPU memory
Validated Configuration: Scale-out AI performance of design is validated by NVIDIA and each storage vendor

DDN A3I + DGX BasePOD Digital Whitepaper

Dive deep into the architecture and performance of a DGX BasePOD configuration with DDN A3I storage. Read detailed performance and throughput metrics.

Learn More

Get in touch

Microway Sales Engineer for Assistance

Get DGX POD Quote