Contact me immediately if you encounter problems!

All Categories
wentian wa7780 g3 ai large model training 7u rack server 2 socket 8 gpu server-0

Thinksystem Server

Home >  Products >  Server >  Thinksystem Server

WenTian WA7780 G3 AI Large Model Training 7U Rack Server 2-socket 8 GPU Server

  • Overview
  • AI Rack Server
  • Related Products

WenTian WA7780 G3 AI Server

Lenovo Wentian WA7780 G3 is a server specifically designed for AI large model training. With the advent of the general artificial intelligence era, the parameter quantity of large model training has surpassed the trillion-level threshold, and AI computing power has become a crucial requirement for enterprises to train, develop, and utilize artificial intelligence. Lenovo Wentian WA7780 G3 is here to break through the dilemma of AI computing power supply, providing powerful computing support for general large model training.

Lenovo Wentian WA7780 G3 AI large model training server is based on the fourth and fifth generations of Intel® Xeon® Scalable Processors, equipped with eight NVIDIA HGX™ NVLink® GPUs.

Lenovo Wentian WA7780 G3 AI Large Model Training Server is designed to meet the demands of high-speed data communication between GPU servers in large-scale AI model training scenarios. It supports various external network connection solutions including IB and RoCE. The server can support up to 8 RDMA high-speed network cards, providing an aggregated bandwidth of 3.2 Tb/s. This fully satisfies the communication requirements across nodes during large-scale model training parallel computing. Meanwhile, in terms of local storage design, the server adopts GPU Direct Storage (GDS) technology. Through GDS, GPUs and NVMe achieve fast and low-latency interconnection, which can effectively improve training data loading speed and enhance large model training performance.

Tech Specs

From Factor 7U Rack Server
Processor Up to 2 4th or 5th generation Intel Xeon Scalable Processors
Memory 32 DDR5 DIMMs, with each CPU supporting 8 channels; the maximum frequency is 5600 MHz for 1DPC and 4800 MHz for 2DPC.
Drive Bays

8x front-panel hot-swappable 2.5-inch storage device slots

-PCIe direct: 8x NVMe SSD

2x rear-panel hot-swappable 2.5-inch storage device slots

-CPU2 direct: 2x NVMe/SATA (PCH)/SAS* SSD (*SAS support requires optional HBA/RAID card)

2x built-in M.2 SSD

GPU Support NVIDIA HGX 8-GPU NVLink
Network Interface Integrated 2 x 10GbE RJ45 LAN Module (Intel x710 Based-T)
RAID SATA with RAID (0,1,5,10); Universal components including standard and internal PCIe RAID/HBA adapters
Power 3+3 3000W 80 PLUS Titanium Power Supply

WA7780 G3 (1).jpgWenTian WA7780 G3 (4).jpgWenTian WA7780 G3 (3).jpgWenTian WA7780 G3 (1).jpgWenTian WA7780 G3 (2).jpg

Get a Free Quote

Our representative will contact you soon.
Email
Name
Company Name
Message
0/1000

Get a Free Quote

Our representative will contact you soon.
Email
Name
Company Name
Message
0/1000

Get a Free Quote

Our representative will contact you soon.
Email
Name
Company Name
Message
0/1000