NVIDIA GH100 Graphics Processing Unit (GPU)

Specs based on NVIDIA H100 SXM 80GB

NVIDIA Hopper 5nm

Most Powerful Variant

Typical Memory Size

Typical Memory Bus

GH100 GPU

Manufacturer

NVIDIA

Release Date

March 22nd, 2022

Architecture

Hopper

Fabrication Process

5 nm

Die Size

814 mm²

Transistors Count

80B

Transistors Density

98.3M ^TRAN/_mm²

Tensor Cores

528

CUDAs

16896

SM

132

GPCs

8

Boost Clock

1775 MHz

Memory Size

81920 MB

Memory Type

HBM3

Memory Bus Width

5120-bit

Memory Bandwidth

3,072.0 GB/s

SPFP Performance

60 TFLOPS

Performance per W

85.7 GFLOPS/W

Performance per mm²

73.7 GFLOPS/mm²

Interface

SXM 5.0

Typical Thermal Design Power

700 W

DirectX

-

Vulkan

-

OpenGL

-

OpenCL

-

Pictures

GH100 GPU Cards4

	Model	GPU Model	Unified Cores	Boost Clock	Memory Clock	Memory Config.
	NVIDIA H100 PCIe 120GB	GH100	TBC	TBC	TBC	TBC
	NVIDIA H100 SXM 80GB	GH100	16896	1775 MHz	5 Gbps	80 GB 5120b
	NVIDIA H100 CNX	GH100	14592	1775 MHz	3 Gbps	80 GB 5120b
	NVIDIA H100 PCIe 80GB	GH100	14592	1775 MHz	3 Gbps	80 GB 5120b