second-generation Transformer Engine
GPTKB entity
Statements (16)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:technology
|
| gptkbp:announced |
2022
|
| gptkbp:developedBy |
gptkb:NVIDIA
|
| gptkbp:enables |
efficient inference for generative AI
faster large language model training |
| gptkbp:feature |
FP8 precision support
automated mixed-precision computation dynamic range management |
| https://www.w3.org/2000/01/rdf-schema#label |
second-generation Transformer Engine
|
| gptkbp:integratesWith |
gptkb:NVIDIA_Hopper_architecture
|
| gptkbp:purpose |
accelerate AI transformer model training and inference
|
| gptkbp:successor |
gptkb:first-generation_Transformer_Engine
|
| gptkbp:usedIn |
gptkb:NVIDIA_H100
gptkb:NVIDIA_GH200 |
| gptkbp:bfsParent |
gptkb:Nvidia_Blackwell
|
| gptkbp:bfsLayer |
6
|