跳到主要内容Skip to main content

← 返回硬件算力← Back to Compute

加速卡Accelerators

NVIDIA H100 NVLNVIDIA H100 NVL

回答「H100 NVL 的 94GB HBM3、3.9TB/s、FP16 1,671 TF 与 350–400W」。1,671 TF 是带稀疏的 FP16 Tensor Core 数字;dense 需按官方脚注折算。
不主张:不是 H100 SXM(80GB / 3.35TB/s / 700W);不是 H200;不是整机 8 卡数字。
Answers H100 NVL's 94GB HBM3, 3.9TB/s, 1,671 TF FP16 and 350–400W. 1,671 TF is the sparse FP16 Tensor Core figure; dense needs the official footnote conversion.
It does not claim: this is not H100 SXM (80GB / 3.35TB/s / 700W); it is not H200; it is not an 8-GPU system figure.

关键参数Key specs

HBM3
94 GB
带宽
3.9 TB/s
FP16 Tensor(稀疏)
1,671 TFLOPS
FP8 Tensor(稀疏)
3,341 TFLOPS
功耗
350–400 W
形态
PCIe 双槽

官方原文怎么说What the source says

GPU Memory | 80GB | 94GB
1 https://www.nvidia.com/en-us/data-center/h100/
GPU Memory Bandwidth | 3.35TB/s | 3.9TB/s
2 https://www.nvidia.com/en-us/data-center/h100/
FP16 Tensor Core | 1,979 teraFLOPS | 1,671 teraFLOPS
3 https://www.nvidia.com/en-us/data-center/h100/
350-400W (configurable)
4 https://www.nvidia.com/en-us/data-center/h100/

资料与许可Source and licence

资料时间Source checked
2026-11-19
许可 / 类型Licence / kind
official-spec-page
许可说明Licence evidence
https://www.nvidia.com/en-us/data-center/h100/
类目Category
加速卡Accelerators

去官网看原页Open the official page →

← 返回硬件算力← Back to Compute