Tensor Core acceleration for all popular convolutions including 2D, 3D, Grouped, Depth-wise separable, and Dilated with NHWC and NCHW inputs and outputs.

For access to NVIDIA optimized deep learning framework containers that have cuDNN integrated into frameworks, visit NVIDIA GPU CLOUD to learn more and get started.ĭownload cuDNN Developer Guide Forums Latest Release Notes cuDNN accelerates widely used deep learning frameworks, including Caffe2, Chainer, Keras, MATLAB, MxNet, PaddlePaddle, PyTorch, and TensorFlow. It allows them to focus on training neural networks and developing software applications rather than spending time on low-level GPU performance tuning. cuDNN provides highly tuned implementations for standard routines such as forward and backward convolution, pooling, normalization, and activation layers.ĭeep learning researchers and framework developers worldwide rely on cuDNN for high-performance GPU acceleration.

The NVIDIA CUDA® Deep Neural Network library (cuDNN) is a GPU-accelerated library of primitives for deep neural networks.

Deep Work instal