Statements (22)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:distributed_computing_backend
|
| gptkbp:availableOn |
gptkb:Linux
|
| gptkbp:developedBy |
gptkb:NVIDIA
|
| gptkbp:firstReleased |
2016
|
| gptkbp:fullName |
NVIDIA Collective Communications Library backend
|
| gptkbp:language |
gptkb:C++
|
| gptkbp:openSource |
true
|
| gptkbp:optimizedFor |
gptkb:NVIDIA_GPUs
|
| gptkbp:supports |
gptkb:broadcaster
multi-GPU communication all-gather all-reduce collective operations reduce-scatter |
| gptkbp:usedBy |
gptkb:TensorFlow
gptkb:PyTorch |
| gptkbp:usedIn |
deep learning
distributed training |
| gptkbp:website |
https://developer.nvidia.com/nccl
|
| gptkbp:bfsParent |
gptkb:PyTorch_Distributed
|
| gptkbp:bfsLayer |
7
|
| https://www.w3.org/2000/01/rdf-schema#label |
NCCL backend
|