
NVIDIA H100 NVL 94GB
NVIDIA H100 NVL 94GB
Specification H100 NVL
GPU Architecture NVIDIA Hopper
GPU NVIDIA H100 Tensor Core GPU
GPU Memory 94GB HBM3
Memory Interface 5120-bit
Memory Bandwidth 3.9 TB/s
CUDA Cores 16,896
Tensor Cores 528, 4th Generation
RT Cores 132, 2nd Generation
FP64 Performance 34 TFLOPS
FP32 Performance 67 TFLOPS
FP64 Tensor Core Performance 67 TFLOPS
TF32 Tensor Core Performance 989 TFLOPS
BF16 Tensor Core Performance 1,979 TFLOPS
FP16 Tensor Core Performance 1,979 TFLOPS
FP8 Tensor Core Performance 3,958 TFLOPS
INT8 Tensor Performance 3,958 TOPS
System Interface PCIe 5.0 x16
NVLink 600 GB/s
TDP 350W
Form Factor Dual-slot PCIe
Cooling Active
Display Outputs None
ECC Yes
Memory Type HBM3
Typical Configuration 2 × H100 NVL GPUs
Combined GPU Memory 188GB HBM3
Target Market Data Center / AI / HPC
Main Applications LLM Inference, AI Training, HPC, Scientific Computing
Specification H100 NVL
GPU Architecture NVIDIA Hopper
GPU NVIDIA H100 Tensor Core GPU
GPU Memory 94GB HBM3
Memory Interface 5120-bit
Memory Bandwidth 3.9 TB/s
CUDA Cores 16,896
Tensor Cores 528, 4th Generation
RT Cores 132, 2nd Generation
FP64 Performance 34 TFLOPS
FP32 Performance 67 TFLOPS
FP64 Tensor Core Performance 67 TFLOPS
TF32 Tensor Core Performance 989 TFLOPS
BF16 Tensor Core Performance 1,979 TFLOPS
FP16 Tensor Core Performance 1,979 TFLOPS
FP8 Tensor Core Performance 3,958 TFLOPS
INT8 Tensor Performance 3,958 TOPS
System Interface PCIe 5.0 x16
NVLink 600 GB/s
TDP 350W
Form Factor Dual-slot PCIe
Cooling Active
Display Outputs None
ECC Yes
Memory Type HBM3
Typical Configuration 2 × H100 NVL GPUs
Combined GPU Memory 188GB HBM3
Target Market Data Center / AI / HPC
Main Applications LLM Inference, AI Training, HPC, Scientific Computing
- Product Detail
PREVIOUS:NVIDIA HGX B100
AI Smart Device Home
