|
gptkbp:instance_of
|
gptkb:Library
|
|
gptkbp:available_on
|
gptkb:Linux
gptkb:mac_OS
gptkb:Windows
|
|
gptkbp:designed_for
|
deep learning
|
|
gptkbp:developed_by
|
gptkb:NVIDIA_Corporation
|
|
gptkbp:has
|
gptkb:Documentation
gptkb:API
community support
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
NVIDIA cu DNN
|
|
gptkbp:includes
|
tensor cores
|
|
gptkbp:integrates_with
|
gptkb:cu_BLAS
gptkb:cu_FFT
gptkb:CUDA
cu SPARSE
|
|
gptkbp:is
|
gptkb:fast
cross-platform
flexible
maintained
open source
scalable
efficient
widely used
modular
reliable
used in research
frequently updated
used in production
highly optimized
|
|
gptkbp:is_compatible_with
|
gptkb:NVIDIA_GPUs
|
|
gptkbp:is_optimized_for
|
gptkb:NVIDIA_Ampere_architecture
gptkb:NVIDIA_Turing_architecture
gptkb:NVIDIA_Volta_architecture
|
|
gptkbp:provides
|
performance tuning
multi-GPU support
automatic differentiation
mixed precision training
FP16 support
INT8 support
high-performance primitives
|
|
gptkbp:released_in
|
gptkb:2014
|
|
gptkbp:supports
|
convolutional neural networks
RNNs
GRUs
LSTMs
|
|
gptkbp:used_in
|
gptkb:Tensor_Flow
gptkb:Caffe
gptkb:MXNet
gptkb:Py_Torch
|
|
gptkbp:written_in
|
gptkb:C++
|
|
gptkbp:bfsParent
|
gptkb:CUDA
|
|
gptkbp:bfsLayer
|
4
|