Designed for advanced GPU workloads, the server supports 8x onboard AMD Instinct MI300X GPUs with an OAM form factor., connected via PCIe 5.0 x16 CPU-to-GPU interconnects and AMD Infinity Fabric Link for high-bandwidth, low-latency data transfer. Storage options include 8x front hot-swap 2.5" NVMe drive bays and 1x M.2 NVMe slot on the motherboard. Up to 6TB of memory can be installed across 24x DIMM slots, with memory speeds reaching 4800MT/s on AMD EPYC 9004 processors and up to 6000MT/s on 9005 series CPUs. All components are powered by 4x 5250W redundant (2+2) Titanium level power supplies. Redundancy is based on the configuration and application load.
For expansion, the 4125GS-TNMR2-LCC offers 8x PCIe 5.0 x16 low-profile slots and 2x PCIe 5.0 x16 FHHL slots by default. Administrators have the option of installing 8x PCIe 5.0 x16 low-profile slots and 4x PCIe 5.0 x16 FHHL slots. To manage the GPU server, Supermicro SuperDoctor 5 (SD5) provides real-time monitoring for power consumption, fan speeds, and CPU temperatures. Automated alerts via email or Simple Network Management Protocol (SNMP) can help organizations promptly fix issues. Additional tools such as SuperCloud Composer, Supermicro Server Manager, and the SuperServer Automation Assistant are also offered.
This system can only be purchased with a minimal configuration including 2x CPUs with 48 cores minimum, each, but CPUs with 56 cores are recommended. Also required, at least 1.5TB of memory, 8x AMD instinct MI300X OAM GPUs, 1x storage device and one network interface controller (NIC).
Memory Slots:
Max GPU Count:
Supported GPU:
CPU-GPU Interconnect:
GPU-GPU Interconnect:
Default: Total 8 bays
M.2:
Default:
Option A:
Chipset:
Choose components, get instant pricing, and build the perfect server in minutes.
Fast, simple, and completely transparent!