客服热线：010-62958599

首页 > AI Servers and Accessories > Nvidia GPU > H100
NVIDIA H100 GPU
￥267000

型号：NVIDIA H100 GPU

加入购物车

产品特征

The leap from Ampere to Hopper

4th-Generation Tensor Core and the FP8 Revolution

Confidential Computing and Energy Efficiency Optimization

Performance Improvement and Application Scenarios

  • 商品详情
  • 资料下载(PDF)

1



恒正同创第二次版详情页_03.jpg

The NVIDIA H100 GPU is a Tensor Core GPU based on the NVIDIA Hopper architecture, delivering

unprecedented performance and scalability for a wide range of workloads. It supports the NVIDIA

NVLink system, allowing connection of up to 256 H100 GPUs to accelerate large-scale AI and

high-performance computing (HPC) tasks. Equipped with a dedicated Transformer Engine, the

H100 GPU efficiently processes trillion-parameter language models, boosting AI training and

inference performance.


Brand

NVIDIA

Model

H100 80G PCIE

Part Number

900-21010-0000-000

Product Dimensions

268mmX111mm双槽

Chip Manufacturer

NVIDIA

Chip Model

H100 80G PCIE

Core Bus Width

5120bit

Graphics Card Slot

PCIe 5.0 x16

Interface

PCIe

Video Memory Capacity

80GB

Video Memory Type

HBM2e/3

Memory Clock Frequency

6144-bit

Core Clock Frequency

1000MHz（MHz）

Memory Clock Frequency

1593MHz（MHz）

Stream Processors (SP)

14592Units

3D API

DirectX 12


2.jpg

3-副本.jpg

6.jpg


  • 客服热线：010-62958599
  • Monday to Friday 9:00-18:00
  • 关注恒正：

Copyright © 2009-2023 北京恒正同创科技有限公司 版权所有者 京ICP备19047959号-3

TOP
×