(858) 613-1818Quote list 0

2U four node multi node server

Four independent dual socket nodes in one 2U chassis with shared redundant power, EPYC 9005 or Xeon 6, DDR5 and NVMe, for dense CPU tiers and HPC.

Illustration of a 2U rackmount server
Nodes
4 x hot pluggable dual socket nodes in 2U, 8 CPUs per chassis
CPU
2 x AMD EPYC 9005 or 2 x Intel Xeon 6 per node
Memory
16 or 24 x DDR5 RDIMM per node, up to 1.5 TB per node
Storage
2 x M.2 boot and up to 6 x E3.S or U.2 NVMe per node
Networking
1 x 400 Gb/s ConnectX-7 per node, 2 x 25 GbE per node
Power
Shared redundant PSUs, about 3.5 kW per chassis planning figure

Overview

The 2U four node server packs eight sockets into two rack units by sharing power supplies and fans across four independent dual socket nodes. Each node has its own processors, memory, NVMe and network adapters and its own management controller, so the chassis behaves like four 1U servers at roughly half the rack space and better power efficiency.

It is the usual way we deliver the CPU tier of large GPU clusters, where the sizing rule of six threads and 24 GB per GPU calls for many nodes, and it is the standard building block for CPU based HPC, CFD and genomics clusters. A 400 Gb/s ConnectX-7 adapter per node joins each one to the same InfiniBand or Ethernet fabric as the GPU servers.

Nodes can be ordered with AMD EPYC 9005 or Intel Xeon 6, with DDR5 to 1.5 TB per node and up to six NVMe drives each. Plan for about 3.5 kW per chassis at full load. Burn in runs all four nodes at once and the acceptance report lists each node by serial number.

Use cases

  • CPU tier of GPU clusters
  • HPC and simulation clusters
  • Hyperconverged and container platforms
  • Web scale services

Specifications

Compute
Nodes4 independent nodes, each with its own BMC, hot pluggable from the rear
CPUs per node2 x AMD EPYC 9005 series or 2 x Intel Xeon 6
Memory per node16 or 24 x DDR5 RDIMM, up to 1.5 TB
Chassis totalUp to 1,536 cores and 6 TB memory in 2U with EPYC 9005
Storage
Boot per node2 x M.2 NVMe, mirrored
Data per nodeUp to 6 x E3.S or U.2 NVMe, front hot swap
RAIDSoftware RAID
Networking
Fabric per node1 x 400 Gb/s ConnectX-7 or 200 Gb/s, PCIe Gen5
Frontend per node2 x 25 GbE onboard
ManagementDedicated 1 GbE BMC port per node
Physical
Form factor2U rackmount, air cooled, 4 nodes
Power suppliesShared redundant 3,000 W class PSUs, 200 to 240 V input
Planning powerAbout 3.5 kW per chassis, about 0.9 kW per node

Figures describe the system class. Exact configuration, weights and power draw are confirmed on the quotation.

What's included

  • Rails, power cords and cable management arm
  • Burn in and acceptance test report
  • Operating system imaging and firmware baseline on request
  • 3 year warranty, options to 5 years with advance replacement

Compliance

This product is not a data center GPU system and does not require NVIDIA end user registration. Standard export documentation applies. Read about export compliance.

Compare it with the next class up or down

The nearest classes we quote are the 1U dual socket EPYC 9005 server and the 2U dual socket Xeon 6 server. If neither is quite right, tell us the workload and we specify the system between them.

Talk to an engineerRequest a quotation