oneMKL interfaces are an open-source implementation of the oneMKL Data Parallel C++ (DPC++) interface according to the oneMKL specification. It works with multiple devices (backends) using device-specific libraries underneath.

• Support and Requirements
• Build Setup
• Building with Conan
• Building with CMake
• Project Cleanup
• FAQs
• Legal Information

There are two oneMKL selector layer implementations:

• Run-time dispatching: The application is linked with the oneMKL library and the required backend is loaded at run-time based on device vendor (all libraries should be dynamic).

Example of app.cpp with run-time dispatching:

#include "oneapi/mkl.hpp"

...
cpu_dev = sycl::device(sycl::cpu_selector());
gpu_dev = sycl::device(sycl::gpu_selector());

sycl::queue cpu_queue(cpu_dev);
sycl::queue gpu_queue(gpu_dev);

oneapi::mkl::blas::column_major::gemm(cpu_queue, transA, transB, m, ...);
oneapi::mkl::blas::column_major::gemm(gpu_queue, transA, transB, m, ...);

How to build an application with run-time dispatching:

$> clang++ -fsycl –I$ONEMKL/include app.cpp
$> clang++ -fsycl app.o –L$ONEMKL/lib –lonemkl

• Compile-time dispatching: The application uses a templated backend selector API where the template parameters specify the required backends and third-party libraries and the application is linked with the required oneMKL backend wrapper libraries (libraries can be static or dynamic).

Example of app.cpp with compile-time dispatching:

#include "oneapi/mkl.hpp"

...
cpu_dev = sycl::device(sycl::cpu_selector());
gpu_dev = sycl::device(sycl::gpu_selector());

sycl::queue cpu_queue(cpu_dev);
sycl::queue gpu_queue(gpu_dev);

oneapi::mkl::backend_selector<oneapi::mkl::backend::mklcpu> cpu_selector(cpu_queue);

oneapi::mkl::blas::column_major::gemm(cpu_selector, transA, transB, m, ...);
oneapi::mkl::blas::column_major::gemm(oneapi::mkl::backend_selector<oneapi::mkl::backend::cublas> {gpu_queue}, transA, transB, m, ...);

How to build an application with compile-time dispatching:

$> clang++ -fsycl –I$ONEMKL/include app.cpp
$> clang++ -fsycl app.o –L$ONEMKL/lib –lonemkl_blas_mklcpu –lonemkl_blas_cublas

Supported domains: BLAS, LAPACK, RNG

• CPU
  • Intel Atom(R) Processors
  • Intel(R) Core(TM) Processor Family
  • Intel(R) Xeon(R) Processor Family
• Accelerators
  • Intel(R) Processor Graphics GEN9
  • NVIDIA(R) TITAN RTX(TM) (Linux* only. cuRAND backend tested also with Quadro and A100 GPUs. Not tested with other NVIDIA GPU families and products.)

What should I download?

If Building with Conan, above packages marked as "No" must be installed manually.

If Building with CMake, above packages must be installed manually.

LEGAL NOTICE: By downloading and using this container or script as applicable (the “Software Package”) and the included software or software made available for download, you agree to the terms and conditions of the software license agreements for the Software Package, which may also include notices, disclaimers, or license terms for third party software (together, the “Agreements”) included in this README file.

If the Software Package is installed through a silent install, your download and use of the Software Package indicates your acceptance of the Agreements.

conan-center: https://api.bintray.com/conan/conan/conan-center

conan-community: https://api.bintray.com/conan/conan-community/conan

1. Install Intel(R) oneAPI DPC++ Compiler (select variant as per requirement).

2. Clone this project to <path to onemkl>, where <path to onemkl> is the root directory of this repository.

3. You can Build with Conan to automate the process of getting dependencies or you can download and install the required dependencies manually and Build with CMake directly.

Note: Conan package manager automates the process of getting required packages, so that you do not have to go to different web location and follow different instructions to install them.

** This method currently works on Linux* only **

** Make sure you have completed Build Setup. **

Note: To understand how dependencies are resolved, refer to the Product and Version Information section. For details about Conan package manager, refer to Conan Documentation.

Conan can be installed from pip:

pip3 install conan

Conan stores all files and data in ~/.conan. If you are fine with this behavior, you can skip to Conan Profiles section.

To change this behavior, set the environment variable CONAN_USER_HOME to a path of your choice. A .conan/ directory will be created in this path and future Conan commands will use this directory to find configuration files and download dependent packages. Packages will be downloaded into $CONAN_USER_HOME/data. To change the "/data" part of this directory, refer to the [storage] section of conan.conf file.

To make this setting persistent across terminal sessions, you can add below line to your ~/.bashrc or custom runscript. Refer to Conan Documentation for more details.

export CONAN_USER_HOME=/usr/local/my_workspace/conan_cache

Profiles are a way for Conan to determine a basic environment to use for building a project. This project ships with profiles for:

• Intel(R) oneAPI DPC++ Compiler for x86 CPU and Intel GPU backend: inteldpcpp_lnx

1. Open the profile you wish to use from <path to onemkl>/conan/profiles/ and set COMPILER_PREFIX to the path to the root folder of compiler. The root folder is the one that contains the bin and lib directories. For example, Intel(R) oneAPI DPC++ Compiler root folder for default installation on Linux is /opt/intel/inteloneapi/compiler/<version>/linux. User can define custom path for installing the compiler.

COMPILER_PREFIX=<path to Intel(R) oneAPI DPC++ Compiler>

2. You can customize the [env] section of the profile based on individual requirements.

3. Install configurations for this project:

# Inside <path to onemkl>
$ conan config install conan/

This command installs all contents of <path to onemkl>/conan/, most importantly profiles, to conan default directory.

Note: If you change the profile, you must re-run the above command before you can use the new profile.

1. Out-of-source build

# Inside <path to onemkl>
mkdir build && cd build

2. If you choose to build backends with the Intel(R) oneAPI Math Kernel Library, install the GPG key as mentioned here, https://software.intel.com/en-us/articles/oneapi-repo-instructions#aptpkg

3. Install dependencies

conan install .. --profile <profile_name> --build missing [-o <option1>=<value1>] [-o <option2>=<value2>]

The conan install command downloads and installs all requirements for the oneMKL DPC++ Interfaces project as defined in <path to onemkl>/conanfile.py based on the options passed. It also creates conanbuildinfo.cmake file that contains information about all dependencies and their directories. This file is used in top-level CMakeLists.txt.

-pr | --profile <profile_name> Defines a profile for Conan to use for building the project.

-b | --build <package_name|missing> Tells Conan to build or re-build a specific package. If missing is passed as a value, all missing packages are built. This option is recommended when you build the project for the first time, because it caches required packages. You can skip this option for later use of this command.

4. Build Project

conan build .. [--configure] [--build] [--test]  # Default is all

The conan build command executes the build() procedure from <path to onemkl>/conanfile.py. Since this project uses CMake, you can choose to configure, build, test individually or perform all steps by passing no optional arguments.

5. Optionally, you can also install the package. Similar to cmake --install . --prefix <install_dir>.

conan package .. --build-folder . --install-folder <install_dir>

-bf | --build-folder Tells Conan where to find the built project.

-if | --install-folder Tells Conan where to install the package. It is similar to specifying CMAKE_INSTALL_PREFIX

Note: For a detailed list of commands and options, refer to the Conan Command Reference.

The following options are available to pass on conan install when building the oneMKL library:

• build_shared_libs=[True | False]. Setting it to True enables the building of dynamic libraries. The default value is True.
• target_domains=[<list of values>]. Setting it to blas or any other list of domain(s), enables building of those specific domain(s) only. If not defined, the default value is all supported domains.
• enable_mklcpu_backend=[True | False]. Setting it to True enables the building of oneMKL mklcpu backend. The default value is True.
• enable_mklgpu_backend=[True | False]. Setting it to True enables the building of oneMKL mklgpu backend. The default value is True.
• enable_mklcpu_thread_tbb=[True | False]. Setting it to True enables oneMKL on CPU with TBB threading instead of sequential. The default value is True.

• build_functional_tests=[True | False]. Setting it to True enables the building of functional tests. The default value is True.

• build_doc=[True | False]. Setting it to True enables the building of rst files to generate HTML files for updated documentation. The default value is False.

Note: For a mapping between Conan and CMake options, refer to build options under the CMake section.

# Inside <path to onemkl>
mkdir build && cd build
conan install .. --build missing --profile inteldpcpp_lnx -o build_shared_libs=False
conan build ..

1. Make sure you have completed Build Setup.

2. Build and install all required dependencies.

Then:

• On Linux*

# Inside <path to onemkl>
mkdir build && cd build
export CXX=<path_to_dpcpp_compiler>/bin/dpcpp;
cmake .. [-DMKL_ROOT=<mkl_install_prefix>] \                    # required only if environment variable MKLROOT is not set
         [-DREF_BLAS_ROOT=<reference_blas_install_prefix>] \    # required only for testing
         [-DREF_LAPACK_ROOT=<reference_lapack_install_prefix>]  # required only for testing
cmake --build .
ctest
cmake --install . --prefix <path_to_install_dir>

• On Windows*

# Inside <path to onemkl>
md build && cd build
cmake .. -G Ninja
                  [-DMKL_ROOT=<mkl_install_prefix>] \                    # required only if environment variable MKLROOT is not set
                  [-DREF_BLAS_ROOT=<reference_blas_install_prefix>] \    # required only for testing
                  [-DREF_LAPACK_ROOT=<reference_lapack_install_prefix>]  # required only for testing

ninja 
ctest
cmake --install . --prefix <path_to_install_dir>

All options specified in the Conan section are available to CMake. You can specify these options using -D<cmake_option>=<value>.

The following table provides a detailed mapping of options between Conan and CMake.

Note: build_functional_tests and related CMake option affects all domains at a global scope.

Most use-cases involve building the project without the need to cleanup the build directory. However, if you wish to cleanup the build directory, you can delete the build folder and create a new one. If you wish to cleanup the build files but retain the build configuration, following commands will help you do so. They apply to both Conan and CMake methods of building this project.

# If you use "GNU/Unix Makefiles" for building,
make clean

# If you use "Ninja" for building
ninja -t clean

See CONTRIBUTING for more information.

Distributed under the Apache license 2.0. See [LICENSE](LICENSE) for more

information.

1. What is the difference between the following oneMKL items?
   • The oneAPI Specification for oneMKL
   • The oneAPI Math Kernel Library (oneMKL) Interfaces Project
   • The Intel(R) oneAPI Math Kernel Library (oneMKL) Product

Answer:

• The oneAPI Specification for oneMKL defines the DPC++ interfaces for performance math library functions. The oneMKL specification can evolve faster and more frequently than implementations of the specification.

• The oneAPI Math Kernel Library (oneMKL) Interfaces Project is an open source implementation of the specification. The project goal is to demonstrate how the DPC++ interfaces documented in the oneMKL specification can be implemented for any math library and work for any target hardware. While the implementation provided here may not yet be the full implementation of the specification, the goal is to build it out over time. We encourage the community to contribute to this project and help to extend support to multiple hardware targets and other math libraries.

• The Intel(R) oneAPI Math Kernel Library (oneMKL) product is the Intel product implementation of the specification (with DPC++ interfaces) as well as similar functionality with C and Fortran interfaces, and is provided as part of Intel® oneAPI Base Toolkit. It is highly optimized for Intel CPU and Intel GPU hardware.

1. I am behind a proxy. How can Conan download dependencies from external network?

   • ~/.conan/conan.conf has a [proxies] section where you can add the list of proxies. For details refer to Conan proxy settings.

2. I get an error while installing packages via APT through Conan.

   dpkg: warning: failed to open configuration file '~/.dpkg.cfg' for reading: Permission denied
   Setting up intel-oneapi-mkl-devel (2021.1-408.beta07) ...
   E: Sub-process /usr/bin/dpkg returned an error code (1)
   

   • Although your user session has permissions to install packages via sudo apt, it does not have permissions to update debian package configuration, which throws an error code 1, causing a failure in conan install command.
   • The package is most likely installed correctly and can be verified by:
     1. Running the conan install command again.
     2. Checking /opt/intel/inteloneapi for mkl and/or tbb directories.