The site uses Cookies. When you continue to browse the site, you are agreeing our use of Cookies. Read our privacy policy>

  • Products
  • Solutions
  • Support
  • About Us
Global/English
title
  • meta brain® General Purpose Servers
  • meta brain® Artificial Intelligence Servers
  • meta brain® Edge Computing Servers
  • Storage
  • Edge computing and the Internet of Things
  • Data storage and Management
Home Products meta brain® Artificial Intelligence Servers NF5488A5
NF5488A5
NF5488A5
NF5488A5
NF5488A5
NF5488A5
NF5488A5
NF5488A5
NF5488A5
NF5488A5
NF5488A5
NF5488A5
NF5488A5

NF5488A5

NF5488A5 is an exciting new addition to the comprehensive AI portfolio. Ranking first in single-sever performance in the MLPerf AI benchmarks, it continues to lead the benchmark of AI training and inference performance. In the image classification scenario, NF5488A5 can complete a ResNet50 training task in 27 minutes. In the inference task, 329037 pictures can be classified per second. The high performance of NF5488A5 makes it an ideal choice for a series of AI scenarios such as intelligent image, video and voice processing, financial analysis, and virtual assistants.


Key Features
  • Strong power core
    Strong power core

    2x AMD EPYC processors with TDP of 280W, 4 XGMI links for interconnection

    8x NVIDIA Ampere Tensor Core GPUs with high interconnection speed through NVSwitch

    Outstanding Performance in the MLPerf AI benchmarks


  • Excellent system design
    Excellent system design

    High computing power density with 8x Ampere GPUs and 2x CPUs in a 4U space

    Excellent heat dissipation capabilities to support stable operation under high ambient temperatures

    Intensity and efficiency for large-scale data center and industry scenario deployments


  • High-speed and low-delay communication
    High-speed and low-delay communication

    Innovative interconnection architecture significantly reduces communication delay between the CPU and GPU

    800Gb/s InfiniBand high-speed network supports multi-node linear expansion


  • Software and hardware optimization
    Software and hardware optimization

    Software and hardware co-optimization for reduction of data transmission by 50%

    Innovative multi-card scheduling mechanism for linear expansion of GPU performance


Technical Specifications

Model

NF5488A5

GPU

1* HGX Ampere 8-GPU

Processor

2 x AMD® EPYC® Rome/Milan Processor (Up to 280W TDP)

Memory

32 x DDR4 3200MT/s RDIMMs//LRDIMM

Storage

8 x 2.5" SAS/SATA SSD (up to 4 x NVMe U.2) , 6 x M.2

PCIe

5* PCIe 4.0x16 slots

Front I/O

2 x USB 3.0 port, 1 x VGA port

Rear I/O

2 x RJ45 management port

Network

10G Ethernet Optical Interface (option)

Raid

Support 3508 Raid Card or PCle Raid Card

Management

Built-in Aspeed2500 BMC module, support IPMI, SOL, KVM Over IP, VM, etc.

PSU

3+1 Redundant PSUs (3000W 80Plus platinum)

Cooling

N+1 Redundant hot swap fans

OS

Red Hat Enterprise 7.8 64bit, CentOS 7.8, Ubuntu 18.04 or higher version

Size

4U, W x H x D : 448mm x 175.35mm x 834.4mm

Weight

Net weight: 65kg (gross weight: 97kg)

Environment

 

Working temperature：10℃~35℃;  Storage temperature：-40℃~70℃

Working humidity：10%~80% R.H.；Storage humidity：10%~93% R.H.

The working temperature is 0 ~ 40 ℃ at 0 to 1000 meters (3300 feet);

Operating temperature 5 ~ 32 ℃ from 1000 to 3050 m (10000 feet)


Please enter the SN or MN code of your product to preview or download it.

Verification code：

By clicking Submit,you acknowledge that you agree to comply with all applicable laws and regulations.

Click here to learn how to inquire a SN code.