.rst .pdf

Contents

Welcome to oneAPI

oneAPI is an open, free, and standards-based programming model that provides portability and performance across accelerators and generations of hardware. oneAPI is comprised of the following language and libraries for creating parallel applications:

• DPC++ Language Reference : oneAPI’s core language for programming accelerators and multiprocessors

• oneCCL : Communication primitives for scaling deep learning frameworks across multiple devices

• oneDAL : Algorithms for accelerated data science

• oneDNN : High performance implementations of primitives for deep learning frameworks

• oneDPL : A companion to the oneAPI DPC++/C++ Compiler for programming oneAPI devices with APIs from C++ standard library, Parallel STL, and extensions

• oneMKL : High performance math routines for science, engineering, and financial applications

• oneTBB : Library for adding thread-based parallelism to complex applications on multiprocessors

• oneVPL : Algorithms for accelerated video processing

oneAPI simplifies software development by providing the same languages and programming models across accelerator architectures.

Contribute

We welcome community feedback and contributions to the oneAPI component documentation. Refer to each component’s GitHub repository for the project-specific contribution guidelines:

• DPC++ Language Reference

• oneAPI Collective Communications Library (oneCCL)

• oneAPI Data Analytics Library (oneDAL)

• oneAPI Deep Neural Network Library (oneDNN)

• oneAPI DPC++ Library (oneDPL)

• oneAPI Math Kernel Library (oneMKL)

• oneAPI Threading Building Blocks (oneTBB)

• oneAPI Video Processing Library (oneVPL)

DPC++ Language Reference