NVIDIA DGX H200

Specification NVIDIA DGX H200
Product Name NVIDIA DGX H200
System Type AI / HPC Data Center Server
GPU 8 × NVIDIA H200 Tensor Core GPU
GPU Memory per GPU 141GB HBM3e
Total GPU Memory 1,128GB HBM3e
GPU Architecture NVIDIA Hopper
GPU Memory Bandwidth 4.8 TB/s per GPU
CPU 2 × Intel Xeon Platinum 8480C
CPU Cores 112 cores total
System Memory 2TB DDR5 ECC
NVSwitch 4 × NVIDIA NVSwitch
NVLink 4th Generation NVLink
GPU-to-GPU Bandwidth 900 GB/s
Cluster Network 8 × NVIDIA ConnectX-7
Network Speed Up to 400Gb/s per card
OS Storage 2 × 1.92TB NVMe M.2 SSD, RAID 1
Data Cache 8 × 3.84TB NVMe U.2 SED, RAID 0
System Memory DIMMs 32 × DIMMs
Power Supply 6 × 3.3kW
Form Factor 8U Rackmount
Cooling Air Cooling
BMC 1GbE RJ45
Target Market Data Center / AI / HPC
Main Applications LLM Training, LLM Inference, Generative AI, HPC