Skip to main content

NVIDIA H100 (Hopper)

Product Overview

The NVIDIA H100 is a flagship data center GPU based on the Hopper architecture GH100 chip, released in 2022. As of 2025, it remains the de facto standard for large language model training and inference. It introduces the Transformer Engine (dynamic FP8 precision acceleration) and the DPX instruction set (dynamic programming algorithm acceleration).

Core Specifications

ParameterValue
ArchitectureHopper GH100
Process NodeTSMC 4N (custom 5nm)
Transistor Count80 billion
Memory80 GB HBM3
Memory Bandwidth3.35 TB/s (3,352 GB/s)
CUDA Cores16,896
Tensor Cores528 (4th Gen)
FP3267 TFLOPS
FP6434 TFLOPS (important for HPC)
TF32 Tensor Core989 TFLOPS (sparse)
FP16/BF16 Tensor Core1,979 TFLOPS (sparse)
FP8 Tensor Core3,958 TFLOPS (sparse)
INT8 Tensor Core3,958 TOPS (sparse)
TDP700 W (SXM5)
InterconnectNVLink 4.0 (900 GB/s), PCIe 5.0
MIGUp to 7 instances
Form FactorSXM5 / PCIe 5.0

Vendor Information

ParameterValue
ManufacturerNVIDIA Corporation
Official Websitehttps://www.nvidia.com
Product Pagehttps://www.nvidia.com/en-us/data-center/h100/
ReleaseMarch 2022 GTC
EOLMostly replaced by H200 / Blackwell across channels

Software & Drivers

ResourceLink
Data Center Driverhttps://www.nvidia.com/Download/index.aspx
CUDA Toolkithttps://developer.nvidia.com/cuda-toolkit
TensorRThttps://developer.nvidia.com/tensorrt
NVIDIA AI Enterprisehttps://www.nvidia.com/en-us/data-center/products/ai-enterprise/
NCCL (Multi-GPU Communication)https://developer.nvidia.com/nccl

Key Features

  • Transformer Engine: Automatically switches between FP8 and FP16 to accelerate Transformer training
  • 4th Gen Tensor Cores: Support FP8 (E4M3, E5M2)
  • DPX Instructions: Hardware-accelerated dynamic programming algorithms
  • MIG: Virtualizes a single GPU into up to 7 independent GPU instances
  • NVLink 4.0 + NVSwitch 3.0: 900 GB/s intra-server interconnect

Use Cases

  • LLM training and fine-tuning
  • Recommendation systems and multimodal AI
  • HPC scientific computing
  • Large-scale distributed training clusters