(858) 613-1818Quote list 0

8 GPU RTX 5090 AI server, 4U

Eight GeForce RTX 5090 GPUs with 256 GB of GDDR7 in a 4U air cooled server on dual EPYC 9005 or Xeon 6, for dense batch inference and media generation.

GPU: NVIDIA GeForce RTX 5090

Illustration of a 4U GPU server
GPUs
8 x NVIDIA GeForce RTX 5090, 32 GB GDDR7 each, 256 GB total
CPU
2 x AMD EPYC 9005 or 2 x Intel Xeon 6
Memory
Up to 2 TB DDR5 RDIMM
Storage
2 x M.2 boot, up to 8 x U.2 NVMe
Networking
2 x 25 GbE onboard, optional 200 or 400 Gb/s adapter
Power
Redundant PSUs, about 6.5 kW per node planning figure

Overview

The 8 GPU RTX 5090 server is the dense version of the 4 GPU class: eight GeForce RTX 5090 cards with 256 GB of GDDR7 behind dual AMD EPYC 9005 or Intel Xeon 6 processors in four rack units. It is built for batch inference at the lowest cost per GPU, for media generation farms that run image and video models around the clock, and for render nodes.

Eight independent GPUs with FP4 Tensor cores and 1.8 TB/s of memory bandwidth each hold eight replicas of a model up to the 30B class, or four pairs for models in the 70B class at FP4 split over PCIe. Hardware AV1 encode and decode on every card suits video pipelines. PCIe Gen5 switches give each GPU a full x16 link; there is no NVLink and no ECC on the GPU memory.

GeForce cards are consumer products: NVIDIA does not license the GeForce driver for data center deployment, and the GPU memory is not ECC protected. The server suits buyers who accept those terms for an office server room, a studio or a development cluster, and we ask you to confirm the intended use on the quotation. For ECC, MIG and professional drivers, the 8 GPU PCIe server with RTX PRO 6000 Blackwell Server Edition is the step up.

Use cases

  • Dense batch inference
  • Media generation farms
  • Render nodes
  • Fine tuning small and mid size models

Specifications

Compute
GPUs8 x NVIDIA GeForce RTX 5090, two slot cards, 575 W each
GPU memory32 GB GDDR7 per GPU, 512 bit, 1.8 TB/s, no ECC
GPU interconnectPCIe Gen5 switches, x16 per GPU, no NVLink
CPUs2 x AMD EPYC 9005 series or 2 x Intel Xeon 6
System memory24 x DDR5 RDIMM, up to 2 TB
Storage
Boot2 x M.2 NVMe, mirrored
DataUp to 8 x U.2 NVMe, hot swap
RAIDSoftware RAID, optional hardware RAID or HBA
Networking
Onboard2 x 25 GbE
Optional1 or 2 x 200 or 400 Gb/s ConnectX-7 adapters
ManagementDedicated 1 GbE BMC port, TPM 2.0
Physical
Form factor4U rackmount, air cooled, high static pressure fans
Power suppliesRedundant 3,000 W class PSUs, 200 to 240 V input
Planning powerAbout 6.5 kW per node
Driver and licensingGeForce driver; the buyer confirms the intended use on the quotation

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

What's included

  • Rails and power cords
  • Burn in with all GPUs under sustained load and acceptance report
  • Operating system imaged with NVIDIA drivers and CUDA toolkit 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 4 GPU RTX 5090 AI server, 4U and the 8 GPU RTX PRO 6000 PCIe AI server. If neither is quite right, tell us the workload and we specify the system between them.

Talk to an engineerRequest a quotation