We take cryptocurrency — so the coins you hold can buy real hardware.Pay in crypto, spend it on real hardware. Free insured shipping worldwide over $399. Plain, unbranded boxes, sent from Germany.Free insured shipping over $399. Your rate is locked for 30 minutes once checkout opens.Rate locked 30 minutes. Every item sealed, serial-checked and covered by a 24-month warranty.Sealed, 24-month warranty.

Saved items Sign in

ASRock Rack

ASRock Rack 1U2N2G-ROME/2T 1U dual-node GPU barebone server

SKU CH-GNFB-Y5HCF MPN 1U2N2G-ROME/2T

1U dual-node barebone server for AI/ML/HPC with dual 10G LAN and two GPU slots per node

  • WorkloadAI/ML/HPC
  • CPU TypeSupports AMD EPYC 7003 (with AMD 3D V-Cache Technology)/7002
  • CPU SocketSocket SP3
  • Dual-node density fits two EPYC servers in 1U
  • 1600W redundant power keeps both nodes running under load
  • PCIe 4.0 x16 per node accepts full-height dual-slot GPU
  • Dual 10GbE Intel X550 per node moves dataset traffic fast
  • IPMI remote management controls each node from anywhere

Dual-node GPU server for AI and HPC

The ASRock Rack 1U2N2G-ROME/2T packs two independent compute nodes into a single 1U rackmount chassis. Each node accepts an AMD EPYC 7003 or 7002 series processor in a Socket SP3 interface. The platform targets AI/ML/HPC workloads that benefit from dense GPU acceleration.

Verify CPU cooler clearance and GPU power budget

Each node provides four DDR4 RDIMM or LRDIMM slots in a 1DPC layout and one full-height dual-slot PCIe 4.0 x16 slot for an add-in card. One M.2 socket per node accepts PCIe 4.0 x4 or SATA storage. Dual 10GbE RJ45 ports driven by an Intel X550-AT2 controller handle network traffic. IPMI remote management is included. Confirm that your chosen processor cooler fits the 1U height and that the GPU power draw stays within the shared 1600W redundant supply capacity.

Sustained load depends on airflow and power headroom

The 1+1 1600W CRPS units feed both nodes simultaneously, so continuous full-load operation on both GPUs and CPUs will test the chassis cooling limits. Thermal throttling may appear if inlet temperatures rise or if high-TDP processors are paired with power-hungry accelerators. Plan rack airflow and monitor IPMI sensors during extended training or simulation runs.

Highlights

  • Dual-node density fits two EPYC servers in 1U
  • 1600W redundant power keeps both nodes running under load
  • PCIe 4.0 x16 per node accepts full-height dual-slot GPU
  • Dual 10GbE Intel X550 per node moves dataset traffic fast
  • IPMI remote management controls each node from anywhere

Specifications

WorkloadAI/ML/HPC
CPU TypeSupports AMD EPYC 7003 (with AMD 3D V-Cache Technology)/7002 Series Processors
CPU SocketSocket SP3
Shipping weight20.87 kg
Package size864 × 635 × 254 mm

Questions about this item

Does the barebone include CPUs, memory or GPUs, or must they be purchased separately?

The unit is a barebone chassis with dual nodes, 1+1 1600W CRPS power supplies, dual 10GbE Intel X550-AT2 LAN and IPMI; CPUs, DDR4 RDIMM/LRDIMM modules and full-height dual-slot PCIe 4.0 x16 GPUs are not included.

Which EPYC processor generation should I select for maximum cache per core?

Choose an AMD EPYC 7003 series CPU with 3D V-Cache Technology for the largest L3 cache per core; 7002 series processors are also supported but lack the extra cache layer.

What does the single PCIe 4.0 x16 slot per node mean for GPU installation?

Each node accepts one full-height dual-slot graphics or compute card at PCIe 4.0 x16 bandwidth, so a pair of such cards can be installed across the two nodes.

New to this? Buying server hardware with crypto — what to check, what it costs to pay, and what is never asked for.

Also in this aisle

People compared these

Same shelf, same checkout — eight coins and a 30-minute rate lock.