Description
NVIDIA DGX BasePOD
Built on NVIDIA DGX H200 platform
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
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
DGX H200 Systems
NVIDIA DGX BasePOD can be configured with 2 - 16 DGX System to best suit the usecase.
LEARN MORETB of GPU Memory
18 048 GB of graphics memory can be utilized with 16 DGX systems installed.
LEARN MOREGb InfiniBand Fabric
BasePOD systems utilize NVIDIA SN4600 switches to provide maximum throughput.
LEARN MOREDGX BasePOD Networking
The NVIDIA DGX BasePOD uses QM9700, QM8700, SN4600 and SN2201 switches with NVIDIA ConnectX-7 HCAs to reach up to 400 GbE data transfer between individual systems interconnected with HDR InfiniBand fabric.
Storage Solutions
IBM Storage
IBM FlashSystem storage offers excelling data transfer performane from which AI training benefits. All-flash solutions can deliver great results when paired with NVIDIA DGX BasePOD, minimizing iteration time and improving model inference.
WekaIO
The WEKA Data Platform with NVIDIA DGX BasePOD reference architecture delivers the flexibility to scale resources based on evolving computational needs and provides cost-efficient, streamlined management.
VAST Data
VAST Data was founded on the idea that the future of artificial intelligence must be built upon fast infrastructure that allows for AI engines to process data at any scale. VAST Data storage systems offer full-flash solutions fulfilling these ideas.
NVIDIA DGX BasePOD H200 system diagram
The NVIDIA DGX BasePOD system is interconnected with two layers of NVIDIA QM and SN switches, creating multiple networking routes and levels. Individual DGX H200 modules are connected with high-speed 400Gbps InfiniBand through 4 dedicated lanes, following by connection to management server nodes and storage connected through the SN4600 switches. This creates configurable hirearchy for the system that can be adjusted to the owners needs.
NVIDIA Base Command
NVIDIA Base CommandTM powers the NVIDIA DGXTM platform, enabling organizations to leverage the best of NVIDIA AI innovation. With it, every organization can tap the full potential of their DGX infrastructure with a proven platform that includes AI workflow management, enterprise-grade cluster management, libraries that accelerate compute, storage, and network infrastructure, and system software optimized for running AI workloads.
NVIDIA NGC Platform
NVIDIA NGC, which the BasePOD utilizes, is the portal of enterprise services, software, management tools, and support for end-to-end AI and digital twin workflows. Bring your solutions to market faster with fully managed services, or take advantage of performance-optimized software to build and deploy solutions on your preferred cloud, on-prem, and edge systems.
Language Modelling
Language modeling is a natural language processing (NLP) task that determines the probability of a given sequence of words occurring in a sentence.
HPC
High-performance computing (HPC) is one of the most essential tools fueling the advancement of computational science, and that universe of scientific computing has expanded in all directions.
ASR
Automatic speech recognition (ASR) systems include giving voice commands to an interactive virtual assistant, converting audio to subtitles on an online video, and more.
Image Processing
Image segmentation is the field of image processing that deals with separating an image into multiple subgroups or regions that represent distinctive objects or subparts.
NVIDIA DGX SuperPOD
NVIDIA DGX SuperPOD is an AI data center infrastructure that enables IT to deliver performance—without compromise—for every user and workload. As part of the NVIDIA DGX platform, DGX SuperPOD offers leadership-class accelerated infrastructure and scalable performance for the most challenging AI workloads, with industry-proven results.
Reviews
There are no reviews yet.