(858) 613-1818Quote list 0

2 GPU 2U inference server

Two full power PCIe GPUs, RTX PRO 6000 Blackwell Server Edition, H200 NVL or L40S, in a 2U server with 12 drive bays, for inference and fine tuning.

GPU: NVIDIA RTX PRO 6000 Blackwell Server Edition, H200 NVL or L40S

Illustration of a 2U rackmount server
GPUs
2 x RTX PRO 6000 Blackwell Server Edition (96 GB), H200 NVL (141 GB, NVLink bridge) or L40S (48 GB)
CPU
1 or 2 x AMD EPYC 9005, or 2 x Intel Xeon 6
Memory
Up to 1.5 TB DDR5 RDIMM
Storage
2 x M.2 boot, up to 12 x U.2 or E3.S NVMe
Networking
1 x 200 Gb/s or 400 Gb/s ConnectX-7, 2 x 25 GbE onboard
Power
Redundant PSUs, about 2 kW per node planning figure

Overview

The 2U two GPU server takes full power 600 W cards that the 1U class cannot, so it is the entry point to RTX PRO 6000 Blackwell Server Edition and H200 NVL. Two RTX PRO 6000 cards provide 192 GB of GDDR7 for serving models up to about 70B parameters at FP4, and a bridged pair of H200 NVL provides 282 GB of HBM3e with 900 GB/s between the cards.

Its twelve drive bays make it a natural node for retrieval augmented generation, vector databases and analytics that keep data close to the GPUs. It is also the usual fine tuning box for teams that train adapters on their own data and serve the result from the same machine, and a sensible first server before scaling to the 4 and 8 GPU classes.

Hosts are single or dual AMD EPYC 9005 or dual Intel Xeon 6 with DDR5 to 1.5 TB. One 200 Gb/s or 400 Gb/s ConnectX-7 adapter joins it to a cluster fabric or a high speed storage network. At about 2 kW it runs on 120 V circuits with L40S and on 200 to 240 V with Blackwell cards.

Use cases

  • LLM inference up to 70B parameters
  • RAG and vector search
  • Fine tuning with adapters
  • Analytics and data science
  • Media processing

Specifications

Compute
GPUs2 x NVIDIA RTX PRO 6000 Blackwell Server Edition, H200 NVL or L40S, double width PCIe Gen5
GPU interconnectPCIe Gen5 x16 per GPU; 2 way NVLink bridge with H200 NVL at 900 GB/s
CPUs1 or 2 x AMD EPYC 9005 series, or 2 x Intel Xeon 6
System memory24 x DDR5 RDIMM, up to 1.5 TB
Storage
Boot2 x M.2 NVMe, mirrored
DataUp to 12 x U.2 or E3.S NVMe, hot swap
RAIDSoftware RAID or hardware RAID
Networking
Fabric1 x 200 Gb/s or 400 Gb/s ConnectX-7, InfiniBand or RoCE
Frontend2 x 25 GbE onboard, optional 2 x 100 GbE
ManagementDedicated 1 GbE BMC port, TPM 2.0
Physical
Form factor2U rackmount, air cooled, validated for 600 W GPUs
Power suppliesRedundant 1,600 W class PSUs, 100 to 240 V input
Planning powerAbout 2 kW per node with 600 W GPUs

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
  • NVIDIA end user registration handled with the order

Compliance

Compare it with the next class up or down

The nearest classes we quote are the 2 GPU 1U inference server and the 4 GPU PCIe AI server, 2U. If neither is quite right, tell us the workload and we specify the system between them.

Talk to an engineerRequest a quotation