NVIDIA® CUDA® Deep Neural Network library (cuDNN) is a GPU-accelerated library of primitives for deep neural networks. cuDNN provides highly tuned implementations for standard routines, such as forward and backward convolution, attention, matmul, pooling, and normalization. Download cuDNN Library Download cuDNN Frontend (GitHub) cuDNN is also available to download via one of the package managers below. Quick Install with conda Installs the cuDNN library Quick Pull with Docker Installs the cuDNN lLibrary Quick Install with pip Installs the cuDNN library Installs the cuDNN Frontend API How cuDNN Works Accelerated Llearning: cuDNN provides kernels, targeting Tensor Cores whenever it makes sense, to deliver best- available performance on compute-bound operations. It offers heuristics for choosing the right kernel for a given problem size. Accelerated Llearning: cuDNN provides kernels, targeting Tensor Cores whenever it makes sense, to deliver best- available performance on compute-bound operations. It offers heuristics for choosing the right kernel for a given problem size. Fusion Support: cuDNN supports fusion of compute-bound and memory-bound operations. Common generic fusion patterns are typically implemented by runtime kernel generation. Specialized fusion patterns are optimized with pre-written kernels. Fusion Support: cuDNN supports fusion of compute-bound and memory-bound operations. Common generic fusion patterns are typically implemented by runtime kernel generation. Specialized fusion patterns are optimized with pre-written kernels. Expressive Op Graph API: The user defines computations as a graph of operations on tensors. The cuDNN library has both a direct C API and an open-source C++ frontend for convenience. Most users choose the frontend as their entry point to cuDNN Expressive Op Graph API: The user defines computations as a graph of operations on tensors. The cuDNN library has both a direct C API and an open-source C++ frontend for convenience. Most users choose the frontend as their entry point to cuDNN cuDNN API Code Sample The code performs a batched matrix multiplication with bias using the cuDNN PyTorch integration. Sample Operation Graphs Described by the cuDNN Graph API ConvolutionFwd followed by a DAG with two operations Complete guides on installing and using the cuDNN frontend and cuDNN backend. Frontend Samples Samples illustrate usage of the Python and C++ frontend APIs. Latest Release Blog Learn how to accelerate transformers with scaled dot product attention (SDPA) in cuDNN 9. cuDNN on NVIDIA Blackwell Learn about new/updated APIs of cuDNN pertaining to NVIDIA Blackwell’s microscaling format and how to program against those APIs. Deep Neural Networks Deep learning neural networks span computer vision, conversational AI, and recommendation systems and have led to breakthroughs like autonomous vehicles and intelligent voice assistants. NVIDIA's GPU-accelerated deep learning frameworks speed up training time for these technologies, reducing multi-day sessions to just a few hours. cuDNN supplies foundational libraries for high-performance, low-latency inference for deep neural networks in the cloud, on embedded devices, and in self-driving cars. Accelerated compute-bound operations like attention training/prefill, convolution, and matmul Accelerated compute-bound operations like attention training/prefill, convolution, and matmul Optimized memory-bound operations like attention decode, pooling, softmax, normalization, activation, pointwise, and tensor transformation Optimized memory-bound operations like attention decode, pooling, softmax, normalization, activation, pointwise, and tensor transformation Fusions of compute-bound and memory-bound operations Fusions of compute-bound and memory-bound operations

Runtime fusion engine to generate kernels at runtime for common fusion patterns Runtime fusion engine to generate kernels at runtime for common fusion patterns Optimizations for important specialized patterns like fused attention Optimizations for important specialized patterns like fused attention Heuristics to choose the right implementation for a given problem size Heuristics to choose the right implementation for a given problem size cuDNN Graph API and Fusion The cuDNN Graph API is designed to express common computation patterns in deep learning. A cuDNN graph represents operations as nodes and tensors as edges, similar to a dataflow graph in a typical deep learning framework. Access to the cuDNN Graph API is conveniently available through the Python/C++ Frontend API (recommended) as well as the lower-level C Backend API (for legacy use cases or special cases where Python/C++ isn’t appropriate). Flexible fusions of memory-limited operations into the input and output of matmul and convolution Flexible fusions of memory-limited operations into the input and output of matmul and convolution Specialized fusions for patterns like attention and convolution with normalization Specialized fusions for patterns like attention and convolution with normalization Support for both forward and backward propagation Support for both forward and backward propagation Heuristics for predicting the best implementation for a given problem size Heuristics for predicting the best implementation for a given problem size Open-source Python/C++ Frontend API Open-source Python/C++ Frontend API Serialization and deserialization support Serialization and deserialization support cuDNN Accelerated Frameworks cuDNN accelerates widely used deep learning frameworks, including PyTorch, JAX, Caffe2, Chainer, Keras, MATLAB, MxNet, PaddlePaddle, and TensorFlow. Related Libraries and Software NeMo is an end-to-end cloud-native framework for developers to build, customize, and deploy generative AI models with billions of parameters. NVIDIA TensorRT™ TensorRT is a software development kit for high-performance deep learning inference. NVIDIA Optimized Frameworks Deep learning frameworks offer building blocks for designing, training, and validating deep neural networks through a high-level programming interface. NVIDIA Collective Communication Library NCCL is a communication library for high-bandwidth, low-latency, GPU-accelerated networking. Join the Community Join the NVIDIA Developer Program Accelerate Your Startup NVIDIA believes Trustworthy AI is a shared responsibility, and we have established policies and practices to enable development for a wide array of AI applications. When downloading or using a model in accordance with our terms of service, developers should work with their supporting model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse. Please report security vulnerabilities or NVIDIA AI concerns here . Get Started With cuDNN Today Download cuDNN Library Download cuDNN Frontend (GitHub)