Contents

Build from Source

Download the Source Code

Download oneDNN source code or clone the repository.

git clone https://github.com/oneapi-src/oneDNN.git

Build the Library

Ensure that all software dependencies are in place and have at least the minimal supported version.

The oneDNN build system is based on CMake. Use

  • CMAKE_INSTALL_PREFIX to control the library installation location,

  • CMAKE_BUILD_TYPE to select between build type (Release, Debug, RelWithDebInfo).

  • CMAKE_PREFIX_PATH to specify directories to be searched for the dependencies located at non-standard locations.

See Build Options for detailed description of build-time configuration options.

Linux/macOS

GCC, Clang, or Intel C/C++ Compiler

  • Set up the environment for the compiler

  • Configure CMake and generate makefiles

    mkdir -p build
cd build

# Uncomment the following lines to build with Clang
# export CC=clang
# export CXX=clang++

# Uncomment the following lines to build with Intel C/C++ Compiler
# export CC=icc
# export CXX=icpc
cmake .. <extra build options>

  • Build the library

    make -j

oneAPI DPC++ Compiler

  • Set up the environment for oneAPI DPC++ Compiler. For Intel oneAPI Base Toolkit distribution installed to default location you can do this using setenv.sh script

    source /opt/intel/oneapi/setvars.sh

  • Configure CMake and generate makefiles

    mkdir -p build
cd build

export CC=icx
export CXX=icpx

cmake .. \
          -DDNNL_CPU_RUNTIME=DPCPP
          -DDNNL_GPU_RUNTIME=DPCPP
          <extra build options>

Open-source version of oneAPI DPC++ Compiler may not contain OpenCL runtime. In this case, you can use OPENCLROOT CMake option or environment variable of the same name to specify path to the OpenCL runtime if it is installed in a custom location.

Note

Open-source version of oneAPI DPC++ Compiler does not have the icx/icpx drivers, use clang/clang++ instead.

  • Build the library

    make -j

GCC targeting AArch64 on x64 host

  • Set up the environment for the compiler

  • Configure CMake and generate makefiles

    export CC=aarch64-linux-gnu-gcc
export CXX=aarch64-linux-gnu-g++
cmake .. \
          -DCMAKE_SYSTEM_NAME=Linux \
          -DCMAKE_SYSTEM_PROCESSOR=AARCH64 \
          -DCMAKE_LIBRARY_PATH=/usr/aarch64-linux-gnu/lib \
          <extra build options>

  • Build the library

    make -j

GCC with Arm Compute Library (ACL) on AArch64 host

  • Set up the environment for the compiler

  • Configure CMake and generate makefiles

    export ACL_ROOT_DIR=<path/to/Compute Library>
/cmake .. \
          -DDNNL_AARCH64_USE_ACL=ON \
          <extra build options>

  • Build the library

    make -j

Windows

Microsoft Visual C++ Compiler or Intel C/C++ Compiler

  • Generate a Microsoft Visual Studio solution

    mkdir build
cd build
cmake -G "Visual Studio 16 2019 Win64" ..

    For the solution to use the Intel C++ Compiler, select the corresponding toolchain using the cmake -T switch:

    cmake -G "Visual Studio 16 2019 Win64" -T "Intel C++ Compiler 19.0" ..

  • Build the library

    cmake --build . --config=Release

Note

CMake’s Microsoft Visual Studio generator does not respect CMAKE_BUILD_TYPE option. Solution file supports both Debug and Release builds with Debug being the default. You can choose specific build type with --config option.

Note

You can also open oneDNN.sln to build the project from the Microsoft Visual Studio IDE.

oneAPI DPC++ Compiler

  • Set up the environment for oneAPI DPC++ Compiler. For Intel oneAPI Base Toolkit distribution installed to default location you can do this using setvars.bat script

    "C:\Program Files (x86)\Intel\oneAPI\setvars.bat"

    or open Intel oneAPI Commmand Prompt instead.

  • Download oneAPI Level Zero headers from Github and unpack the archive.

  • Generate Ninja project

    mkdir build
cd build

:: Set C and C++ compilers
set CC=icx
set CXX=icx
cmake .. -G Ninja -DDNNL_CPU_RUNTIME=DPCPP ^
                  -DDNNL_GPU_RUNTIME=DPCPP ^
                  -DCMAKE_PREFIX_PATH=<path to Level Zero headers> ^
                  <extra build options>

Open-source version of oneAPI DPC++ Compiler may not contain OpenCL runtime. In this case, you can use OPENCLROOT CMake option or environment variable of the same name to specify path to the OpenCL runtime if it is installed in a custom location.

Note

Open-source version of oneAPI DPC++ Compiler does not have the icx driver, use clang/clang++ instead.

Note

The only CMake generator that supports oneAPI DPC++ Compiler on Windows is Ninja. CC and CXX variables must be set to clang and clang++ respectively.

  • Build the library

    cmake --build .

Validate the Build

If the library is built for the host system, you can run unit tests using:

ctest

Build documenation

  • Install the requirements

    conda env create -f ../doc/environment.yml
conda activate onednn-doc

  • Build the documentation

    cmake --build . --target doc

Install library

Install the library, headers, and documentation

cmake --build . --target install

The install directory is specified by the CMAKE_INSTALL_PREFIX cmake variable. When installing in the default directory, the above command needs to be run with administrative privileges using sudo on Linux/Mac or a command prompt run as administrator on Windows.

Building and Linking Build Options