NVIDIA H100 PCIe 80GB 
Specification NVIDIA H100 PCIe 80GB
Brand NVIDIA
Product NVIDIA H100 PCIe 80GB Tensor Core GPU
Architecture NVIDIA Hopper
GPU GH100
CUDA Cores 14,592
Tensor Cores 456, 4th Generation
GPU Memory 80GB HBM2e
Memory Interface 5,120-bit
Memory Bandwidth 2,000 GB/s (2 TB/s)
FP64 Performance 26 TFLOPS
FP64 Tensor Core 51 TFLOPS
FP32 Performance 51 TFLOPS
TF32 Tensor Core 756 TFLOPS
BF16 Tensor Core 1,513 TFLOPS
FP16 Tensor Core 1,513 TFLOPS
FP8 Tensor Core 3,026 TFLOPS
INT8 Tensor Core 3,026 TOPS
L2 Cache 50MB
System Interface PCIe 5.0 x16
PCIe Bandwidth 128 GB/s
NVLink 600 GB/s
MIG Up to 7 MIG Instances @ 10GB each
TDP 300–350W configurable
Power Connector 1 × PCIe 16-pin
Form Factor PCIe, Full-Height, Full-Length, Dual-Slot
Cooling Passive
Display Outputs None
ECC Yes
Primary Applications AI Training / AI Inference / HPC / Data Analytics