Statements (24)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
Neural Engine
|
| gptkbp:announced |
2018
|
| gptkbp:architecture |
ASIC
|
| gptkbp:clusterSize |
up to 2048 chips (TPU v3 Pod)
|
| gptkbp:connects |
high-speed interconnect
|
| gptkbp:coolingSystem |
liquid cooling
|
| gptkbp:formFactor |
rack-mounted
|
| https://www.w3.org/2000/01/rdf-schema#label |
TPU v3
|
| gptkbp:location |
gptkb:Google_data_centers
|
| gptkbp:manufacturer |
gptkb:Google
|
| gptkbp:performanceContext |
420 teraflops
|
| gptkbp:powerSource |
450 W
|
| gptkbp:predecessor |
gptkb:TPU_v2
|
| gptkbp:processNode |
16 nm
|
| gptkbp:RAM |
32 GB HBM
|
| gptkbp:successor |
gptkb:TPU_v4
|
| gptkbp:supports |
gptkb:TensorFlow
gptkb:PyTorch |
| gptkbp:usedFor |
gptkb:artificial_intelligence
gptkb:machine_learning |
| gptkbp:usedIn |
gptkb:Google_Cloud_Platform
|
| gptkbp:bfsParent |
gptkb:TPU_v4
gptkb:Neural_Engine |
| gptkbp:bfsLayer |
6
|