bandwagonNF5468G7 bandwagonserver (computer) GPUsServer AI server Training server Inference server AI learning

NF5468G7 is the latest product of Wave Information's NF5468 series that sells well both at home and abroad. It is a new generation of AI servers with excellent multi-arithmetic performance and extreme elasticity of architecture expansion, deploying 8 GPUs of the highest performance in a 4U space, and topology switching can be realized according to application scenarios. Equipped with two Intel 4th/5th generation Xeon scalable processors, it provides up to 112 processor cores, 8TB of system memory and 300TB of local high-speedstockpileIn addition, it is oriented to complex application scenarios such as deep learning, meta-universe, AIGC, AI+Science, etc., and builds the strongest adaptive multivariate arithmetic platform in the era of intelligent computing.

GPU server
Wave NF5468G7 Wave Server GPU Server AI Server Training Server Inference Server AI Learning 2

Leadership Architecture

Newly upgraded PCIE Gen5.0 architecture for full-speed operation of 600W GPUs  

Support for the latest GPU graphics cards 

New CXL1.1 interconnect technology supporting storage-level memory expansion and memory space sharing 

Supports the new 400G NDR IB reticulationand 400G OCP 3.0

superior performance

Two Intel® 4th/5th Generation Xeon® Scalable Processors, TDP 350W 

Built-in 8TB of system memory and 300TB of local high-speed storage to boost data access speeds 

Supports four GPU topologies to flexibly match different AI computing scenarios 

PCIE topology with up to 4x bandwidth increase for high throughput data transfer

Flexible Configuration

Modular design concept supports flexible part-level configuration free combinations 

Powerful multi-arithmetic platform adapted to the latest AI acceleration cards from various brands 

Improved thermal performance by 25%, noise reduction by 5%, no interruption maintenance 

Supports N+N redundant 3000W Platinum/Titanium power supplies for ultimate system stability

Ecological excellence

Rich global AI partners, NVIDIA, AMD, Intel, etc. 

Leading deep learning frameworks, TensorFlow, PyTorch, PaddlePaddle, etc. 

Powerful self-developed AIStation resource platform,algorithmic platformand application optimization services 

Mature metabrain ecosystem, providing industry-wide and scene-wide application algorithms and solution services

technical specification

Product ModelNF5468-M7-A0-R0-00
high degree4U
CPUSupports 2 Intel® 4th/5th Generation Xeon® Scalable Processors, TDP 350W
GPU cardSupports 8 full-height, full-length, double-width PCIe interface GPU cards, while supporting ≥ 4 PCIe 5.0 x16 slots
chipsetsIntel® Emmitsburg PCH C740
random access memory (RAM)Supports 32 DDR5 4800MHz RDIMMs
Built-in PCIE cardSupports up to 2 PCIE 5.0 extensions
Front I/O2 USB ports, 1 VGA port
Rear I/O1 serial port, 1 USB 3.0 interface, 1 RJ45 management port, 1 OCP3.0 network card (support NCSI function)
local storageFront 24 x 2.5 or 12 x 3.5" SAS/SATA hard drives, supports up to 16 NVME or E3.S Built-in 2 x M.2 NVME/SATA SSDs
RAID SupportOptional support for RAID0, 1, 10, 5, 50, 6, 60, etc. Support Cache super capacitor protection, provide RAID state migration, RAID configuration memory
operating systemMicrosoft Windows Sever, Red Hat Enterprise Linux, Ubuntu Linux, CentOS and other major operating systems
System coolingN+1 redundant system cooling fans
power supply4 x 1600W/2000W/2200W/3000W 80Plus Platinum/Titanium PSUs supporting N+N redundancy
case (computer) (lit. machine box)Width 447mm, Height 174.5mm, Depth 850mm
operating temperature5 - 35°C / 41°F - 95°F
full complementary weight≤87kg

WeiboWeChatFacebookCopy link