Server Hardware

High-performance AI infrastructure

  • Krambu HGX B300 DLC
    Applications: AI Training, LLM, Generative AI
    CPU: 2× Intel Xeon® 6776P per node
    Chassis: 8 Compute Racks + 4 Management Racks
    Drive: 8× 3.84 TB NVMe per node
    Cooling: Direct Liquid Cooling
    RAM: 216 TB Total Cluster Memory
    GPU: 576x NVIDIA® B300 SXM
    Network Ports: NVIDIA InfiniBand XDR / Spectrum-4
    Power Supply: Redundant, Rack-Integrated Power