Supermicro SYS-822GS-NB3RT AI Server

Flagship 8U AI Server Supermicro SYS-822GS-NB3RT on NVIDIA HGX B300 Platform

Supermicro SYS-822GS-NB3RT is a flagship 8U AI server from the Supermicro Gold Series line, designed for large-scale AI training and inference of the most complex generative models and large language models (LLMs). Unlike the NBRT variant, which uses B200 chips, this configuration is based on the specialized NVIDIA HGX B300 platform.

The model’s key advantage is 2.3 TB of ultra-fast HBM3e memory (288 GB per GPU). This is critical for retaining the context of ultra-large neural networks within a single node: models that previously required distribution across multiple servers can now be trained and served on a single node without performance loss due to network latency.

Use Cases

The Supermicro SYS-822GS-NB3RT is the optimal solution for organizations requiring:

• Large-scale AI training of large language models (LLMs)
• Inference of the most complex generative models
• Working with ultra-large neural networks requiring massive VRAM
• Scientific computing and demanding HPC workloads
• Building AI factories with 800GbE interconnect
• Real-time computer vision and generative AI workloads

AI Accelerators and Compute Base

• Graphics subsystem. Pre-installed NVIDIA HGX B300 board combining 8 Blackwell (SXM) GPUs. The system delivers a total of 2.3 TB of ultra-fast HBM3e memory (288 GB per GPU), critical for retaining the context of ultra-large neural networks within a single node.
• Inter-chip interconnect. Integrated NVSwitch fabric delivers GPU-to-GPU bandwidth up to 1.8 TB/s over 5th generation NVLink, fully eliminating latency in parallel computations.
• Processors. Support for two Intel Xeon 6 processors (6700 and 6500 series) with performance cores (P-cores) in Socket E2 (LGA-4710). The server is designed for CPUs with TDP up to 350 W.
• Memory. 32 DIMM slots for ECC DDR5 RDIMM server memory. Maximum RAM capacity reaches 4 TB at speeds up to 6400 MT/s (1DPC mode) or up to 8 TB in 2DPC configuration (at speeds up to 6000 MT/s).

Network Fabric and Storage Subsystem

• Cluster network cards. The server design is optimized for building AI factories. The system supports up to eight 800GbE OSFP ports (based on NVIDIA ConnectX-8 SuperNIC), delivering extremely high interconnect speeds and ultra-low cluster latency.
• Expansion interfaces. The internal structure includes up to 8 × PCIe 5.0 x16 Low-Profile (LP) slots and 2 × PCIe 5.0 x16 Full-Height Full-Length (FHHL) slots. The interface panel has two 10GbE RJ45 network ports and an IPMI LAN remote management port.
• Data storage. The front bay holds up to 8 × E1.S NVMe drives with hot-swap support. On the motherboard (X14DBG-LC+) there are 2 × M.2 NVMe slots for mirrored OS boot.

Power and Air Cooling

• Power. The fault-tolerant power subsystem consists of 6 redundant power supplies rated at 6600 W each. The units operate in 3+3 balancing and redundancy scheme and are certified to the Titanium Level (96%+) efficiency standard.
• Cooling. The platform relies on powerful air cooling with front-to-back chassis airflow (CSE-GP807TS-R000NP) through an array of high-performance heavy-duty fans with automatic speed control.

⚠️ Important supply condition. According to strict Supermicro regulations, the platform is shipped to the end customer exclusively as a fully assembled and tested system (Complete System). The minimum starting configuration must include at least 2 processors, 2 TB of RAM, an 8-GPU HGX B300 board, and basic network components.

Why GHI SERVERS

• Direct supply from China. We work with CNY and RUB. Turnkey logistics of server equipment from China, including customs clearance and delivery to our Moscow warehouse.
• Custom configuration. We help select the GPU configuration (NVIDIA HGX B300), Intel Xeon 6 processors, memory capacity (DDR5 RDIMM, 1DPC or 2DPC), ConnectX-8 SuperNIC 800GbE network cards, and power supplies for your task and budget.
• Moscow warehouse. Prompt replacement of defective units and warranty service without long downtime.
• 20+ years of expertise. Authorized Supermicro partner since 2005. Pre-sales engineering, configuration assistance, and post-sales support.

Service and Support

• 24/7 technical support
• Pre-sales engineering and configuration assistance
• OEM warranty and full provenance documentation
• Installation, setup, and deployment of server equipment
• Component replacement and warranty service