Statements (25)
Predicate | Object |
---|---|
gptkbp:instanceOf |
Neural Engine
|
gptkbp:application |
AI inference
deep learning training |
gptkbp:architecture |
custom AI accelerator
|
gptkbp:connects |
gptkb:Ethernet
gptkb:RoCE_v2 |
gptkbp:countryOfOrigin |
gptkb:Israel
|
https://www.w3.org/2000/01/rdf-schema#label |
Gaudi2 AI processor
|
gptkbp:manufacturer |
gptkb:Habana_Labs
|
gptkbp:memoryType |
gptkb:HBM2E
|
gptkbp:notableFeature |
high throughput for AI workloads
integrated Ethernet interfaces optimized for large language models scalable multi-chip support |
gptkbp:parentCompany |
gptkb:Intel
|
gptkbp:predecessor |
gptkb:Gaudi_AI_processor
|
gptkbp:processNode |
7nm
|
gptkbp:RAM |
96 GB
|
gptkbp:releaseYear |
2022
|
gptkbp:supports |
gptkb:TensorFlow
gptkb:PyTorch |
gptkbp:usedIn |
gptkb:Amazon_EC2_DL1_instances
gptkb:Supermicro_AI_servers |
gptkbp:bfsParent |
gptkb:Gaudi_AI_processor
|
gptkbp:bfsLayer |
6
|