(858) 613-1818Quote list 0

High frequency single socket server

Single socket server with the highest clock EPYC 9005 or Xeon 6 parts, 8 to 32 cores at 5 GHz class boost, for latency sensitive and licensed software.

Illustration of a 1U rackmount server
CPU
1 x AMD EPYC 9005 high frequency model or Intel Xeon 6, 8 to 32 cores, 5 GHz class boost
Memory
12 x DDR5 RDIMM, up to 768 GB, highest supported speed
Storage
2 x M.2 boot, up to 10 x E3.S or U.2 NVMe
Networking
2 x 25 GbE onboard, 1 x 100 GbE or 200 Gb/s adapter, low latency tuning
Use
Per core licensed databases, trading, simulation pre and post processing, game servers
Power
Redundant PSUs, about 0.6 kW per node planning figure

Overview

Not every workload wants more cores. Databases licensed per core, trading and market data systems, single threaded simulation steps, CAD batch jobs and game servers all run faster on fewer, faster cores with a large cache. This server uses the high frequency models of AMD EPYC 9005 and Intel Xeon 6, with 8 to 32 cores boosting to the 5 GHz class.

In AI environments it serves as the preprocessing and postprocessing node for simulation workflows, the license constrained database beside an inference service, or the latency critical application server that calls the GPU cluster. Memory runs at the highest supported DDR5 speed, and the optional 100 GbE or 200 Gb/s adapter can be tuned for kernel bypass networking.

Storage is mirrored M.2 boot plus up to ten NVMe drives, and the node fits in one rack unit at about 0.6 kW. We can tune BIOS settings for deterministic latency, disable power saving states and document the resulting configuration in the acceptance report.

Use cases

  • Per core licensed databases
  • Low latency trading and market data
  • Simulation pre and post processing
  • Game and application servers

Specifications

Compute
CPU1 x AMD EPYC 9005 F series class or Intel Xeon 6 high frequency SKU, 8 to 32 cores
ClockUp to 5 GHz class boost, large L3 cache per core
System memory12 x DDR5 RDIMM, up to 768 GB at the highest supported data rate
Storage
Boot2 x M.2 NVMe, mirrored
DataUp to 10 x E3.S or U.2 NVMe, hot swap
RAIDSoftware RAID or hardware RAID
Networking
Frontend2 x 25 GbE onboard
Optional1 x 100 GbE or 200 Gb/s ConnectX-7 with kernel bypass support
ManagementDedicated 1 GbE BMC port, TPM 2.0
Physical
Form factor1U rackmount, air cooled
Power suppliesRedundant 800 W class PSUs, 100 to 240 V input
Planning powerAbout 0.6 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
  • BIOS tuning profile documented on request

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 EPYC 9005 control plane server and the 1U dual socket EPYC 9005 server. If neither is quite right, tell us the workload and we specify the system between them.

Talk to an engineerRequest a quotation