Statements (33)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
Neural Engine
|
| gptkbp:architecture |
custom AI processor
|
| gptkbp:competitor |
gptkb:NVIDIA_A100
gptkb:NVIDIA_H100 AMD Instinct MI200 |
| gptkbp:connects |
gptkb:RoCE
gptkb:Ethernet |
| gptkbp:formFactor |
PCIe card
OAM module |
| https://www.w3.org/2000/01/rdf-schema#label |
Gaudi2
|
| gptkbp:manufacturer |
gptkb:Habana_Labs
|
| gptkbp:memoryBusWidth |
2.45 TB/s
|
| gptkbp:memoryType |
gptkb:HBM2E
|
| gptkbp:numberOfAIcores |
24
|
| gptkbp:numberOfMediaEngines |
3
|
| gptkbp:numberOfTensorProcessors |
24
|
| gptkbp:officialWebsite |
https://habana.ai/products/gaudi2/
|
| gptkbp:parentCompany |
gptkb:Intel
|
| gptkbp:platform |
gptkb:TensorFlow
gptkb:PyTorch gptkb:ONNX |
| gptkbp:powerSource |
600W
|
| gptkbp:predecessor |
gptkb:Gaudi
|
| gptkbp:processNode |
7nm
|
| gptkbp:RAM |
96GB
|
| gptkbp:releaseYear |
2022
|
| gptkbp:supports |
deep learning inference
deep learning training |
| gptkbp:targetMarket |
data centers
AI workloads |
| gptkbp:technology |
SynapseAI
|
| gptkbp:bfsParent |
gptkb:Intel_Habana
|
| gptkbp:bfsLayer |
7
|