Drop-in Acceleration on GPUs with Libraries
Access the massively parallel power of a GPU without having to write the GPU code yourself. Libraries provide highly-optimized algorithms and functions you can incorporate into your new or existing applications. Many of the GPU-accelerated libraries are designed to very easily replace existing CPU libraries, minimizing the impacts on existing code.
- Make sure you have an understanding of what GPU Computing is.
- Try CUDA by taking a self-paced lab on nvidia.qwiklab.com. These labs only require a supported web browser and a network that allows Web Sockets. Click here to verify that your network & system support Web Sockets in section "Web Sockets (Port 80)", all check marks should be green.
Download and install the CUDA Toolkit.
You can watch a quick how-to video for Windows showing this process:
- Also see the Getting Started Guides.
- You can watch a quick how-to video for Windows showing this process:
Look through the available GPU accelerated libraries and find one that provides functionality you can use.
- If you are working in C++, you should definitely check out the Thrust parallel template library.
Use the following resources to learn how to add library based GPU acceleration to your applications:
If you use the FFTW API, NVIDIA provides a drop-in replacement with CUFFT.
- Look through the CUDA library code samples that come installed with the CUDA Toolkit.
- If you use the FFTW API, NVIDIA provides a drop-in replacement with CUFFT.
- Browse and ask questions on stackoverflow.com or NVIDIA’s DevTalk forum.
- Learn more by:
The CUDA Toolkit is a free download from NVIDIA and is supported on Windows, Mac, and most standard Linux distributions. Contained within this toolkit are the following libraries:
- CUBLAS – an implementation of BLAS (Basic Linear Algebra Subprograms).
- CUFFT – a Fast Fourier Transform library with support for the FFTW API.
- CURAND – provides facilities that focus on the simple and efficient generation of high-quality pseudorandom and quasirandom numbers.
- CUSPARSE – contains a set of basic linear algebra subroutines used for handling sparse matrices.
- NPP – focuses on imaging and video processing and is widely applicable for developers in these areas.
- In addition to the libraries provided by NVIDIA, there are a large number of other GPU accelerated libraries available.