Home › Servers › Supermicro

Supermicro GPU SuperServer 8U HGX AI SuperServer

CurrentServers8U rack
Supermicro GPU SuperServer 8U HGX AI SuperSe8U rack32 DIMMs

The Supermicro 8U HGX AI SuperServer platform is engineered for large language model (LLM) training and petascale artificial intelligence workloads. Accommodating an 8-GPU NVIDIA HGX baseboard interconnected via NVSwitch, the system delivers high inter-GPU bisection bandwidth combined with dual host CPUs, up to 32 DIMM slots, and dedicated high-speed InfiniBand/RoCE network adapters.

Buy Supermicro GPU SuperServer 8U HGX AI SuperServerSell yours

Specifications

BrandSupermicro
FamilyGPU SuperServer
Form factor8U rack
ReleasedNov 2020
Form factor8U rack
Sockets2
Processor familyIntel Xeon Scalable / AMD EPYC
Memory slots32
Max memory8TB
Gpu support8-way NVIDIA HGX H100 / A100 SXM
Power supplies6x 3000W Redundant Titanium

Works together

How the Supermicro GPU SuperServer 8U HGX AI SuperServer connects to servers, storage and the network: 5 ways, with the parts and products on each path.

Fibre Channel SANiSCSI and NVMe/TCP over EthernetDirect-attached SASFile storage (NFS and SMB)GPU cluster fabric

Fibre Channel SAN

Match the speed: 16Gb and 32Gb FC HBAs, switches and array ports work together, and each generation also runs one or two speeds below its own (a 32Gb port runs at 16Gb and 8Gb).

Switches

Storage

Optics

Fibre Channel SAN: how it fits together

iSCSI and NVMe/TCP over Ethernet

Keep the speed the same at all three points: a 25GbE server, a 25GbE switch port and a 25GbE array port. Mixed speeds work but the slowest link sets the pace.

Adapters for the server (check the fit list)

Switches

Storage

Optics

iSCSI and NVMe/TCP over Ethernet: how it fits together

Direct-attached SAS

12Gb SAS uses Mini-SAS HD (SFF-8644) connectors; older 6Gb gear uses Mini-SAS (SFF-8088). Adapter cables exist, but the link runs at the slower speed.

Adapters for the server (check the fit list)

Storage

Direct-attached SAS: how it fits together

File storage (NFS and SMB)

Front-end ports face the clients; back-end ports (on scale-out systems) connect the nodes to each other and need their own switch pair.

Adapters for the server (check the fit list)

Switches

Storage

Optics

File storage (NFS and SMB): how it fits together

GPU cluster fabric

InfiniBand adapters and switches must match the generation: HDR is 200Gb, NDR is 400Gb. VPI adapters such as ConnectX-6 and ConnectX-7 can run either InfiniBand or Ethernet.

GPUs

Adapters for the server (check the fit list)

Switches

GPU cluster fabric: how it fits together

Plan and size it

Check compatibility with Supermicro

Before you buy, confirm the exact configuration against the vendor's own lists: firmware levels, operating systems, adapters and optics that Supermicro has tested with the Supermicro GPU SuperServer 8U HGX AI SuperServer.

Equivalent models from other brands

FAQ

What GPU interconnect is utilized in the 8U system?

The 8U chassis integrates NVIDIA HGX baseboards where eight SXM GPUs communicate over full NVLink mesh topology with NVSwitch interconnects.

What power infrastructure is required for an 8U HGX system?

The system utilizes six redundant 3000W Titanium-level power supplies operating on 200-240V high-line circuits to deliver up to 10.2kW of conditioned power.

Buying or selling 8U HGX AI SuperServer?

New, refurbished or used. Tell us the quantity and condition and we connect you with a vetted reseller in your region, usually within one working day.

Buy 8U HGX AI SuperServerSell yours

Sources