Statements (31)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:collective_communications_library
|
| gptkbp:abbreviation |
gptkb:NVIDIA_Collective_Communications_Library
|
| gptkbp:communicationBackendFor |
distributed deep learning frameworks
|
| gptkbp:developer |
gptkb:NVIDIA
|
| gptkbp:enables |
gptkb:broadcaster
reduce all-gather all-reduce collective operations reduce-scatter |
| gptkbp:firstReleased |
2016
|
| gptkbp:integratesWith |
gptkb:TensorFlow
gptkb:MXNet gptkb:PyTorch gptkb:Horovod |
| gptkbp:latestReleaseVersion |
2024-03-19
2.19.3 |
| gptkbp:license |
gptkb:BSD_3-Clause_License
|
| gptkbp:operatingSystem |
gptkb:Linux
|
| gptkbp:optimizedFor |
gptkb:NVIDIA_GPUs
|
| gptkbp:programmingLanguage |
gptkb:CUDA
gptkb:C++ |
| gptkbp:supports |
multi-GPU communication
multi-node communication |
| gptkbp:usedFor |
deep learning
high-performance computing |
| gptkbp:website |
https://developer.nvidia.com/nccl
|
| gptkbp:bfsParent |
gptkb:Nvidia_H100
gptkb:NVIDIA_H100 |
| gptkbp:bfsLayer |
7
|
| https://www.w3.org/2000/01/rdf-schema#label |
NVIDIA NCCL
|