NVIDIA H100 Tensor Core GPU

NEW

An Order-of-Magnitude Leap for Accelerated Computing

Tap into unprecedented performance, scalability, and security for every workload with the NVIDIA® H100 Tensor Core GPU. With NVIDIA NVLink® Switch System, up to 256 H100s can be connected to accelerate exascale workloads, along with a dedicated Transformer Engine to solve trillion-parameter language models. H100’s combined technology innovations can speed up large language models by an incredible 30X over the previous generation to deliver industry-leading conversational AI.

Product Update 19-April-2022: Shipping date has not been been announced. XENON will update this page and take pre-orders as soon as shipping dates and prices have been finalised.

XENON NVIDIA DGX H100 PCIE
Specifications
NVIDIA H100 Tensor Core GPU PCIe
Specifications

FP64

NVIDIA H100 Tensor Core GPU PCIe

24 teraFLOPS

Specifications

FP64 Tensor Core

NVIDIA H100 Tensor Core GPU PCIe

48 teraFLOPS

Specifications

FP32

NVIDIA H100 Tensor Core GPU PCIe

48 teraFLOPS

Specifications

TF32 Tensor Core

NVIDIA H100 Tensor Core GPU PCIe

800 teraFLOPS* | 400 teraFLOPS

Specifications

BFLOAT16 Tensor Core

NVIDIA H100 Tensor Core GPU PCIe

1,600 teraFLOPS* | 800 teraFLOPS

Specifications

FP16 Tensor Core

NVIDIA H100 Tensor Core GPU PCIe

1,600 teraFLOPS* | 800 teraFLOPS

Specifications

FP8 Tensor Core

NVIDIA H100 Tensor Core GPU PCIe

3,200 teraFLOPS* | 1,600 teraFLOPS

Specifications

INT8 Tensor Core

NVIDIA H100 Tensor Core GPU PCIe

3,200 TOPS* | 1,600 TOPS

Specifications

GPU memory

NVIDIA H100 Tensor Core GPU PCIe

80GB

Specifications

GPU memory bandwidth

NVIDIA H100 Tensor Core GPU PCIe

2TB/s

Specifications

Decoders

NVIDIA H100 Tensor Core GPU PCIe

  • 7 NVDEC
  • 7 JPEG
Specifications

Max thermal design power (TDP)

NVIDIA H100 Tensor Core GPU PCIe

350W

Specifications

Multi-Instance GPUs

NVIDIA H100 Tensor Core GPU PCIe

Up to 7 MIGS @ 10GB each

Specifications

Form factor

NVIDIA H100 Tensor Core GPU PCIe

PCIe

Specifications

Interconnect

NVIDIA H100 Tensor Core GPU PCIe

  • NVLINK: 600GB/s
  • PCIe Gen5: 128GB/s
Specifications

Server options

NVIDIA H100 Tensor Core GPU PCIe

Partner and NVIDIA-Certified Systems with 1–8 GPUs

Quick Quote Request

  • This field is for validation purposes and should be left unchanged.

Get a Quote Talk to a Solutions Architect