Hardware & Equipment
GP812.png
video
product name

GPU Server for AI Workloads

型號: GP812

產品: General-purpose server for AI applications

特點:

產品描述

Supports 2 x 4th Gen / 5th Gen Intel ® XEON ® Scalable processors, TDP 350W  

Supports up to 8 double-width GPUs

Supports 32 DDR5 memory slots up to 5600MHz with support for RDIMM and 3DS RDIMM with a maximum of 8TB memory

Supports up to 12 standard PCIe slots

Supports 8 GPUs + 1 PCIe 5.0 x16 + 2 PCIe 5.0 x8

Supports 8 GPUs + 4 PCIe 5.0 x8

Supports 12 x 3.5"/2.5" hard drives, optional 8 x 3.5"/2.5" hard drives supporting up to 4 x U.2 NVMe SSDs

Supports 2 x M.2 ( SATA 3.0 / PCIe 4.0 x4, 2280 and 22110 )

1 x 1Gb RJ45 dedicated management network port

Supports 1 x OCP3.0 network card, optional PCIe 5.0 x8 or x16 (pass-through configuration supports PCIe 5.0 x8 only ), supports NCSI

Optional 4 x 1GbE / 2 x 10Gb SFP+ / 4 x 10Gb SFP+ / 2 x 25Gb SFP28, supports other network cards compliant with the OCP 3.0 standard

Front: 2 x USB 3.0 and 1 x VGA port

Rear: 1 serial port, 2 x USB 3.0 ports, 1 x VGA port, 1 x RJ45 management port

8 x 8056 high-speed fans, hot-swappable, N+1 redundant

Optional rear fan module supported

Configurable with 4 x CRPS power supply modules, hot-swappable, supporting 2+2 / 3+1 / 1+1 redundant modes

Optional 2000W / 2400W / 2700W, Platinum power supply

Integrated BMC management chip AST2600, supports IPMI 2.0, Redfish, SOL, KVM, virtual media and more

Optional TPM/TCM security module, chassis intrusion detection, lockable chassis top cover (tool-less)

Supports Microsoft Windows Server, Red Hat Enterprise Linux, SUSE Linux Enterprise Server, CentOS, Ubuntu, Debian and other mainstream operating systems