AI Smart Device Home

Products

NVIDIA HGX B200

NVIDIA HGX B200 8-GPU 
Specification NVIDIA HGX B200
Product Name NVIDIA HGX B200
Platform Type Accelerated Computing Platform
GPU 8 × NVIDIA B200 Tensor Core GPU
GPU Architecture NVIDIA Blackwell
GPU Form Factor SXM6
GPU Memory per GPU 180GB HBM3e
Total GPU Memory 1.44TB HBM3e
Memory Bandwidth per GPU 8TB/s
Aggregate Memory Bandwidth 64TB/s
GPU Interconnect NVIDIA NVLink + NVSwitch
NVLink Generation 5th Generation NVLink
NVLink Bandwidth per GPU 1.8TB/s
Aggregate NVLink Bandwidth 14.4TB/s
NVSwitch 5th Generation NVIDIA NVSwitch
FP4 Tensor Performance 144 PFLOPS
FP8 / FP6 Tensor Performance 72 PFLOPS
INT8 Tensor Performance 72 POPS
FP16 / BF16 Tensor Performance 36 PFLOPS
TF32 Tensor Performance 18 PFLOPS
FP32 Performance 600 TFLOPS
FP64 Performance 296 TFLOPS
GPU TDP Up to 1,000W per GPU
Baseboard Weight ~32kg
Cooling Air / Liquid Cooling
Platform Form Factor HGX GPU Baseboard
Target Market AI / HPC / Data Center
Typical Integration OEM / ODM Server
  • Product Detail

PREVIOUS:NVIDIA HGX B300

NEXT:NVIDIA HGX B100

Leave a Reply

+852 4748 9818

84798770582

yba960270@gmail.com

Leave a message