Product

NF5468M5
4U 16GPU High Performance Server for Large-Scale AI Deployment

Key Features:

  • High Performance for Al Model Training
  • High Efficiency for Al Edge Computing
  • High Reliability for more Al Application Scenarios

Overview

NF5468M5 (AGX-V) is designed for large-scale deployment in data center and optimization of edge computing. Equipped with the advanced Intel Xeon Skylake processor, it supports 16 NVDIA GPUs in a compact 4U chassis, making it an ideal choice for Al Cloud, security, finance, communication, healthcare and other scenarios.

Features

High Performance for Al Model Training

  • Supports 8 global leading NVDIA V100 32G GPU.
  • Integrates 4 cutting-edge technologies: NVLink 2.0, RDMA, GPUDricet, 32G HBM2.
  • Offers highest computing performance and lowest communication latency.

High Efficiency for Al Edge Computing

  • Supports 16 high efficient Tesla P4 GPU and 24x 3.5 HDD.
  • Capable to process 300 streams 1080P video in video structuring application, it is well able to extend the term of validity of data storage to 9 times of world standard.

High Reliability for more Al Application Scenarios

  • Supports over 10 kinds of GPU topology, enabling easy switching without opening the chassis.
  • CPUs are interconnected via 3 UPI, providing industry leading communication bandwidth.
  • Supports 4 2200W 80PLUS Platinum power supply modules and 2+2 redundant strategy.
  • Suitable for all sorts of Al application scenarios.

 

Model NF5468M5
GPU Support Supports 8 * V100, P100, P40 or 16* P4 GPU
Support NVlink 2.0 GPU
Processors 2* Intel Xeon Skylake Gold/Platinum
Memory 24* 2666MHz slot,support DDR4 ECC
Storage 24* 2.5/3.5 HDD(8* NVMe SSD)+ 2* M.2 SSD HDD
Support RAID 0/1/10/5/50/6/60
I/O PCIe Support * 20* PCI-E3.0
Network Dual 2* 10GE Ports
Power 4* 1600W/2000W/2200W 80PLUS

Interested in Learning More?