Properties (48)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:Graphics_Processing_Unit
|
| gptkbp:application |
computer vision
image recognition natural language processing speech recognition recommendation systems |
| gptkbp:architect |
gptkb:application-specific_integrated_circuit_(ASIC)
|
| gptkbp:availableIn |
gptkb:Google_Cloud_Platform
|
| gptkbp:benefits |
lower costs
optimized for TensorFlow better performance than GPUs faster training times |
| gptkbp:collaboratesWith |
gptkb:Google_Research
gptkb:Google_AI gptkb:Google_Brain |
| gptkbp:designedBy |
gptkb:Google
|
| gptkbp:equipment |
memory
control unit interconnect vector unit matrix unit |
| gptkbp:features |
scalability
high energy efficiency customized architecture |
| gptkbp:firstIntroduced |
2016
|
| gptkbp:generator |
second generation
third generation fourth generation |
| gptkbp:hasDepartment |
not suitable for all workloads
less flexible than general-purpose processors requires specific software stack limited_to_Google_Cloud |
| https://www.w3.org/2000/01/rdf-schema#label |
TPU
|
| gptkbp:language |
TensorFlow
|
| gptkbp:market |
artificial intelligence
cloud computing data centers high-performance computing |
| gptkbp:performance |
low latency
high throughput |
| gptkbp:successor |
gptkb:TPU_v4
gptkb:TPU_v3 TPU v2 |
| gptkbp:supports |
deep learning
matrix multiplication neural networks |
| gptkbp:type |
hardware accelerator
|
| gptkbp:usedFor |
machine learning
|