forked from oneapi-src/oneDNN
-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Showing
50 changed files
with
4,437 additions
and
1,492 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,93 @@ | ||
| Build from Source {#dev_guide_build} | ||
| ==================================== | ||
|
|
||
| ## Download the Source Code | ||
|
|
||
| Download [Intel MKL-DNN source code](https://github.com/intel/mkl-dnn/archive/master.zip) | ||
| or clone [the repository](https://github.com/intel/mkl-dnn.git). | ||
|
|
||
| ~~~sh | ||
| git clone https://github.com/intel/mkl-dnn.git | ||
| ~~~ | ||
|
|
||
| ## Build the Library | ||
|
|
||
| Ensure that all software dependencies are in place and have at least the | ||
| minimal supported version. | ||
|
|
||
| The Intel MKL-DNN 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`). | ||
|
|
||
| See @ref dev_guide_build_options for detailed description of build-time | ||
| configuration options. | ||
|
|
||
| ### Linux/macOS | ||
|
|
||
| - Generate makefile | ||
| ~~~sh | ||
| mkdir -p build && cd build && cmake .. | ||
| ~~~ | ||
|
|
||
| - Build the library | ||
| ~~~sh | ||
| make -j | ||
| ~~~ | ||
|
|
||
| - Build the documentation | ||
| ~~~sh | ||
| make doc | ||
| ~~~ | ||
|
|
||
| - Install the library, headers, and documentation: | ||
| ~~~sh | ||
| make install | ||
| ~~~ | ||
|
|
||
| ### Windows | ||
|
|
||
| - Generate a Microsoft Visual Studio solution: | ||
| ~~~bat | ||
| mkdir build && cd build && cmake -G "Visual Studio 15 2017 Win64" .. | ||
| ~~~ | ||
| For the solution to use the Intel C++ Compiler, select the corresponding | ||
| toolchain using the cmake `-T` switch: | ||
| ~~~bat | ||
| cmake -G "Visual Studio 15 2017 Win64" -T "Intel C++ Compiler 19.0" .. | ||
| ~~~ | ||
|
|
||
| - Build the library | ||
| ~~~bat | ||
| cmake --build . | ||
| ~~~ | ||
| You can also use the `msbuild` command-line tool directly (here | ||
| `/p:Configuration` selects the build configuration which can be different from | ||
| the one specified in `CMAKE_BUILD_TYPE`, and `/m` enables a parallel build): | ||
| ~~~bat | ||
| msbuild "Intel(R) MKL-DNN.sln" /p:Configuration=Release /m | ||
| ~~~ | ||
|
|
||
| - Build the documentation | ||
| ~~~bat | ||
| cmake --build . --target DOC | ||
| ~~~ | ||
|
|
||
| - Install the library, headers, and documentation: | ||
| ~~~bat | ||
| cmake --build . --target INSTALL | ||
| ~~~ | ||
|
|
||
| ## Validate the Build | ||
|
|
||
| Run unit tests: | ||
|
|
||
| ~~~sh | ||
| ctest | ||
| ~~~ | ||
|
|
||
| -------- | ||
|
|
||
| [Legal information](@ref legal_information) |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,145 @@ | ||
| Build Options {#dev_guide_build_options} | ||
| ==================================== | ||
|
|
||
| ## Configuring the Build | ||
|
|
||
| Intel MKL-DNN supports the following build-time options. | ||
|
|
||
| | Option | Supported values (defaults in bold) | Description | ||
| | :--- | :--- | :--- | ||
| | MKLDNN_LIBRARY_TYPE | **SHARED**, STATIC | Defines the resulting library type | ||
| | MKLDNN_THREADING | **OMP**, OMP:INTEL, OMP:COMP, TBB | Defines the threading type | ||
| | MKLDNN_USE_MKL | **DEF**, NONE, ML, FULL, FULL:STATIC | Defines the binary dependency on Intel MKL | ||
| | MKLDNN_GPU_BACKEND | **NONE**, OPENCL | Defines the backend for GPU engine | ||
| | MKLDNN_BUILD_EXAMPLES | **ON**, OFF | Controls building the examples | ||
| | MKLDNN_BUILD_TESTS | **ON**, OFF | Controls building the tests | ||
| | MKLDNN_ARCH_OPT_FLAGS | *compiler flags* | Specifies compiler optimization flags (see warning note below) | ||
| | MKLDNN_ENABLE_JIT_PROFILING | **ON**, OFF | Enables integration with Intel(R) VTune(TM) Amplifier | ||
|
|
||
| ## Targeting Specific Architecture | ||
|
|
||
| Intel MKL-DNN uses JIT code generation to implement most of its functionality | ||
| and will choose the best code based on detected processor features. However, | ||
| some Intel MKL-DNN functionality will still benefit from targeting a specific | ||
| processor architecture at build time. You can use `MKLDNN_ARCH_OPT_FLAGS` CMake | ||
| option for this. | ||
|
|
||
| For Intel(R) C++ Compilers, the default option is `-xHOST`, which instructs | ||
| the compiler to generate the code for the architecture of the processor where | ||
| the build is occurring. This option would not allow you to run the library on | ||
| older processor architectures. | ||
|
|
||
| For GNU\* Compiler 5.0 and newer, the default options are `-march=native | ||
| -mtune=native`. | ||
|
|
||
| @warning | ||
| While use of `MKLDNN_ARCH_OPT_FLAGS` option gives better performance, the | ||
| resulting library can be run only on systems that have instruction set | ||
| compatible with the target instruction set. Therefore, `ARCH_OPT_FLAGS` | ||
| should be set to an empty string (`""`) if the resulting library needs to be | ||
| portable. | ||
|
|
||
| ## GPU support (experimental) | ||
|
|
||
| To enable GPU support in Intel MKL-DNN you need to specify the GPU backend | ||
| to use. Currently the only supported one is implemented using OpenCL\* and | ||
| requires Intel(R) SDK for OpenCL\* applications. You can explicitly specify | ||
| the pass to the SDK using `-DOPENCLROOT` cmake option. | ||
|
|
||
| ~~~sh | ||
| cmake -DMKLDNN_GPU_BACKEND=OPENCL -DOPENCLROOT=/path/to/opencl/sdk .. | ||
| ~~~ | ||
|
|
||
| ## Threading | ||
|
|
||
| Intel MKL-DNN can use the OpenMP or TBB threading runtimes. OpenMP threading | ||
| is the default build mode and is recommended for the best performance. TBB | ||
| support is experimental. This behavior is controlled by the `MKLDNN_THREADING` | ||
| CMake option. | ||
|
|
||
| ### OpenMP | ||
|
|
||
| Intel MKL-DNN can use the Intel, GNU, or Clang OpenMP runtimes. Because | ||
| different OpenMP runtimes may not be binary-compatible, it's important to | ||
| ensure that only one OpenMP runtime is used throughout the application. Having | ||
| more than one OpenMP runtime linked to an executable may lead to undefined | ||
| behavior including incorrect results or crashes. | ||
|
|
||
| The Intel MKL-DNN library linked with Intel MKL (full or small libraries | ||
| version) link to the Intel OpenMP runtime included with the Intel MKL. The | ||
| Intel OpenMP runtime is binary compatible with the GNU OpenMP and Clang OpenMP | ||
| runtimes and is recommended for the best performance results. | ||
|
|
||
| If Intel MKL-DNN library is built standalone, it will link to the OpenMP | ||
| library supplied by the compiler. This means that there would be no conflicts | ||
| as long as both the library and the application use the same or compatible | ||
| compilers. | ||
|
|
||
| ### TBB | ||
|
|
||
| TBB support is experimental. | ||
|
|
||
| To build Intel MKL-DNN with TBB support, set the `TBBROOT` environmental | ||
| variable to point to the TBB installation path or pass the path directly to | ||
| cmake: | ||
|
|
||
| ~~~sh | ||
| $ cmake -DTBBROOT=/opt/intel/path/tbb .. | ||
| ~~~ | ||
|
|
||
| Intel MKL-DNN has limited optimizations for Intel TBB and has some functional | ||
| limitations if built with Intel TBB. | ||
|
|
||
| Functional limitations: | ||
| * Winograd convolution algorithm is not supported. | ||
|
|
||
| The following primitives have lower performance compared to OpenMP (mostly due | ||
| to limited parallelism): | ||
| * Batch normalization, | ||
| * Convolution backward by weights, | ||
| * Inner product, | ||
| * `mkldnn_*gemm()`. | ||
|
|
||
| ## Linking to Intel(R) MKL | ||
|
|
||
| Intel MKL-DNN can be configured to use Intel MKL via the `MKLDNN_USE_MKL` | ||
| CMake option. Using Intel MKL may improve performance on older platforms not | ||
| supported by the matrix-matrix multiplication routine implemented in Intel | ||
| MKL-DNN itself. | ||
|
|
||
| If you choose to build Intel MKL-DNN with the binary dependency, download the | ||
| Intel MKL small libraries using the provided script: | ||
|
|
||
| * Linux/macOS | ||
| ~~~sh | ||
| cd scripts && ./prepare_mkl.sh && cd .. | ||
| ~~~ | ||
|
|
||
| * Windows | ||
| ~~~bat | ||
| cd scripts && call prepare_mkl.bat && cd .. | ||
| ~~~ | ||
|
|
||
| You can also download the small libraries from | ||
| [GitHub release section](https://github.com/intel/mkl-dnn/releases). You need | ||
| to unpack the archive in the `external` directory of the repository root or | ||
| set `MKLROOT` environment variable to point to the installation location of | ||
| the small libraries of full Intel MKL library. | ||
|
|
||
| @note | ||
| Intel MKL small libraries are currently only compatible with OpenMP threading. | ||
| You need the full Intel MKL library to use Intel MKL and TBB threading. Using | ||
| Intel MKL or Intel MKL small libraries will introduce additional runtime | ||
| dependencies. For additional information, refer to the Intel MKL | ||
| [system requirements](https://software.intel.com/en-us/articles/intel-math-kernel-library-intel-mkl-2019-system-requirements). | ||
|
|
||
| @warning | ||
| If the library is built with TBB threading, the user is expected to set the | ||
| `MKL_THREADING_LAYER` environment variable to either `tbb` or `sequential` in | ||
| order to force Intel MKL to use Intel TBB for parallelization or to be | ||
| sequential, respectively. Without this setting, Intel MKL (the `mkl_rt` | ||
| library) uses the OpenMP threading by default. | ||
|
|
||
| -------- | ||
|
|
||
| [Legal information](@ref legal_information) |
Oops, something went wrong.