SupermicroAS-1115CS-TNR-01-G2

A 64-core AMD EPYC 9555P at 3.2 GHz, 384 GB of DDR5-6400 and two dual-port 200G NVIDIA ConnectX-7 cards in a 1U CloudDC with ten 2.5-inch bays.

Ask for price and lead timeSee the configuration
Supermicro AS-1115CS-TNR-01-G2Rack 1U · Gold Series G2

The exact configuration

Official datasheet

Front view diagram: hover over each component.
Processor1× AMD EPYC 9555P64 cores · 3.2 GHz
Memory384 GBDDR5-6400
Storage1× 960 GB U.2 NVMe
Network2× NVIDIA ConnectX-7 VPI dual 200G QSFP561× quad 1 GbE RJ45
Form factorRack 1U
Warranty
3 years parts and labour
On-site engineer
Optional: next business day, for 3 years
Software
Supermicro Data Center Management Suite, licence per node

Metadata and serving for AI storage

When many GPUs read from a parallel file system at once, the bottleneck is usually the flood of small metadata operations, and this is the node to handle them inside an AI storage cluster: four 200G ports to talk to the GPUs and the data nodes, and 64 high-frequency cores. The ConnectX-7 cards are VPI, which means each port can run as Ethernet or as InfiniBand.

Outside that role it is not the best buy. A single 960 GB NVMe drive and 6 GB of memory per core make it a poor fit for general virtualisation, and for general compute the AS-1126HS-TN-01 is a better match or, if an entry-level machine will do, its sibling the -02. Before ordering, you need to know which storage system it will run and over what network, because the port mode and the cabling depend on that.

What we do

We set it up as a Ceph metadata server or gateway, with MDS (the CephFS metadata server) or RGW (the gateway that provides S3 access to Ceph), in clusters that serve data to GPUs. It ships with RHEL or Ubuntu, the ConnectX-7 drivers and RDMA networking (direct access to another machine's memory without going through its CPU) configured end to end, from the node to the switch.

Compared with the rest of the family

ModelWhat it hasWhen to choose it
AS-1126HS-TN-01-G22x AMD EPYC 9355 · 1.5 TBTwo high-frequency EPYC in just 1U.
AS-2115HS-TNR-01-G21x AMD EPYC 9655P · 1.5 TB96 cores, 1.5 TB and 200 GbE networking with a DPU.
AS-2115HS-TNR-02-G21x AMD EPYC 9655P · 768 GBThe same 96 cores with faster memory and 25 GbE.
AS-2126HS-TN-01-G22x AMD EPYC 9965 · 1.5 TB384 cores to consolidate, with little memory per core.
SYS-112H-TN-02-G21x Intel Xeon 6731P · 1 TBOne Xeon, 32 GB per core and 4×25 GbE.
SYS-222H-TN-01-G22x Intel Xeon 6740P · 1 TBTwo Xeon in 2U with spare slots and GPU space.
AS-2115GT-HNTR-01-G21x AMD EPYC 9534 · 512 GBFour 64-core servers in 2U, built for clustering.
AS-1115CS-TNR-02-G21x AMD EPYC 9115 · 64 GBEntry-level EPYC 9005 with room to grow inside.
SYS-122C-TN-01-G22x Intel Xeon 6736P · 1 TBTwo Xeon 6 in 1U, for many identical nodes.
AS-3015MR-H8TNR-01-G21x AMD EPYC 4565P · 64 GBEight small, fast nodes in 3U, one per customer.
AS-1015A-MT-01-G21x AMD EPYC 4564P · 32 GBRemote office in 1U and 43 cm deep.

See all rack servers

What the delivery includes

From the factory

  • Configuration assembled and tested by Supermicro
  • Warranty: 3 years parts and labour

From us

  • Rack mounting
  • Remote management (BMC) and up-to-date firmware
  • Operating system and the stack that runs on top
  • Testing with your workload before it goes into production

If you need it

  • 24×7 on-call support with SLA
  • On-site engineer next business day for the full 3 years

How we build it

What runs on top of these servers, explained in detail:

Ceph for AI and HPCAI inference with Ceph, OpenStack and Kubernetes

Support and training for your team

Independent support from the people who installed itDocker, Podman and Kubernetes courseHPC trainingRed Hat OpenShift training

Questions about the AS-1115CS-TNR-01-G2

How does it differ from the AS-1115CS-TNR-02-G2?

They share the chassis and board, and little else. The -01 carries a 64-core EPYC 9555P, 384 GB and two 200G ConnectX-7 cards; the -02, a 16-core EPYC 9115, 64 GB and 10 GbE. They are machines for opposite jobs: the -01 for the network of an AI storage system and the -02 as an entry-level server.

Is there room for data drives?

The chassis has ten 2.5-inch front bays for NVMe, SAS or SATA, and the Gold configuration uses only one, with a 960 GB NVMe drive. If the storage design calls for metadata on local NVMe, nine bays remain free.

Should I set the ports to Ethernet or InfiniBand?

It depends on the network the GPUs and data nodes already use. If the GPU cluster runs on InfiniBand, this node should speak InfiniBand to avoid adding a gateway in between; if it runs on Ethernet with RoCE (RDMA over Ethernet), we leave them on Ethernet. Each port is configured separately, so mixed layouts are possible too.

Is the AS-1115CS-TNR-01-G2 right for what you're building?

Tell us about the workload and the data centre it's going into, and we'll come back with the price, a confirmed lead time and what it would take to get it running.

Ask for price and lead timeWhatsApp+34 91 198 02 43