Server Platform Specs: NVIDIA_DGX_B200
Server and compute platform specifications for AI/HPC systems — NVIDIA DGX A100/H100/H200/B200/GB200 NVL72, NVIDIA DGX SuperPOD GB200, HPE ProLiant DL380 Gen11, and HPE ProLiant DL380a Gen11 GPU server. Includes GPU/CPU configuration, memory, storage, networking, power, physical dimensions, and thermal operating range.
| platform | spec category | ai performance desc | cpu base freq ghz (GHz) | cpu boost freq ghz (GHz) | cpu cores total | cpu count | cpu model | depth mm (mm) | gpu count | gpu memory each gb (GB) | gpu memory total gb (GB) | gpu model | height mm (mm) | max power kw (kW) | network mgmt desc | network scale out desc | network scale up desc | nvswitch count | operating temp max c (degC) | operating temp min c (degC) | rack units (U) | storage capacity tb (TB) | storage internal desc | storage os desc | system memory gb (GB) | system memory max gb (GB) | width mm (mm) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| NVIDIA_DGX_B200 | compute | 72 petaFLOPS FP8 training / 144 petaFLOPS FP4 inference | 2.1 | 4 | 112 | 2 | Intel Xeon Platinum 8570 (dual) | 897 | 8 | 180 | 1,440 | NVIDIA Blackwell B200 GPU | 444 | 14.3 | 10Gb/s onboard NIC (RJ45); 100Gb/s dual-port Ethernet NIC; BMC (RJ45) | 4x OSFP ports serving 8x single-port ConnectX-7 VPI; up to 400Gb/s InfiniBand/Ethernet | 2x dual-port QSFP112 BlueField-3 DPU; up to 400Gb/s InfiniBand/Ethernet | 2 | 30 | 5 | 10 | 30.72 | 8x 3.84TB NVMe U.2 | 2x 1.9TB NVMe M.2 | 2,048 | 4,096 | 482 |
This is a sample. Search the full server platform specs dataset and 350+ more by creating a free account and accessing the database.
Get started for FREE