Eight Blackwell GPUs on the HGX B300 baseboard with 288 GB HBM3e per GPU and 2,304 GB total GPU memory. Connected via NVLink 5 for full all-to-all bandwidth within the node.
Dual Granite Rapids processors with up to 4 TB DDR5 system memory. Enough host compute, I/O bandwidth, and memory capacity to keep all eight GPUs saturated.
Choose between air-cooling and direct liquid-cooled configurations based on your facility capabilities and density targets. Liquid cooling unlocks higher sustained performance at lower operating cost.
Built on Supermicro's enterprise server platform with a track record across thousands of GPU deployments in production environments worldwide.
The Supermicro SYS-822GB-TNR3 is one of two HGX B300 systems available through Arc Compute. Both share the same NVIDIA HGX B300 8-GPU baseboard and NVLink 5 interconnect. The differences are in chassis design, cooling implementation, and system-level integration.
If you are evaluating both options, our team can help you determine which system fits your workload, facility, and deployment requirements.
Every cluster is designed around your workloads, your facility, and your operational model.
Systems are validated, burned in, and shipped with full deployment support.
Hands-on experience designing and deploying high-density GPU infrastructure.
From planning and procurement through deployment, operations, and scaling.
