Skip to content

GPU libraries

The list of GPU libraries and packages provided below are by no means exhaustive, and the world of GPU-accelerated computing is continually evolving. Users are encouraged to conduct their own due diligence, explore additional packages, and stay updated with the latest developments in the GPU computing ecosystem to best suit their specific requirements and project needs.


  1. CUDA Toolkit - The official parallel computing platform and API developed by NVIDIA for GPU acceleration.

  2. CuPy - An open-source GPU-accelerated array library compatible with NumPy.

  3. PyTorch - A popular deep learning framework with GPU support for defining and training neural networks.

  4. TensorFlow - A widely-used deep learning framework for defining and training machine learning models on GPUs.

  5. scikit-cuda - A library providing Python interfaces to various CUDA libraries and functions for GPU acceleration.

  6. RAPIDS - A suite of GPU-accelerated data science libraries, including cuDF, cuML, and cuGraph, developed by NVIDIA.

  7. Numba - A Just-In-Time (JIT) compiler for Python that can generate optimized machine code for CPU and GPU execution.

  8. Theano: (No longer actively developed) - An early deep learning framework that supported GPU acceleration. Many of its concepts have influenced other frameworks.

  9. MXNet - A deep learning framework known for its flexibility and efficiency in training neural networks on GPUs.

  10. Cupy-cuBLAS - An extension of CuPy that provides GPU-accelerated linear algebra operations using the cuBLAS library.

Note: Make sure to check the official websites or repositories of these libraries for the latest installation instructions and documentation.


  1. gputools - Provides GPU-based tools for data manipulation, including matrix operations and basic statistics.

  2. gpuR - Offers GPU support for matrix operations and linear algebra in R.

  3. RapidsR - An R interface to the RAPIDS suite of GPU-accelerated data science libraries developed by NVIDIA, including cuDF and cuML.

  4. mxnet - The R interface to the MXNet deep learning framework, which allows you to train and deploy neural networks on GPUs.

  5. tensorflow - The R interface to TensorFlow, a popular deep learning framework with GPU support.

  6. cudaBayesreg - A package for Bayesian regression analysis with GPU support.

  7. gpuMagic - A package for general-purpose GPU computing in R.

  8. H2O4GPU - A GPU-accelerated machine learning library for R, offering a wide range of machine learning algorithms optimized for GPUs.

  9. Rtorch - An R interface to PyTorch, a popular deep learning framework, allowing you to create and train neural networks on GPUs.

  10. gmatrix - Offers GPU support for matrix operations and linear algebra in R.

Note: Some of these packages may require specific GPU hardware and dependencies. Be sure to check the official documentation and package repositories for installation instructions and system requirements.

You can click on the provided links or search for the package names on CRAN or GitHub to find more information about each package and how to install and use them in your R projects.