TensorFlow windows builds with AVX/AVX2 extensions
Checkout the releases section for the python wheels.
| TF Version | Python Version | Instruction set | GPU Enabled | Link |
|---|---|---|---|---|
| 1.8.0 | 3.5.3 | AVX2 | tensorflow-1.8.0-cp35-cp35m-win_amd64.whl | |
| 1.7.1 | 3.5.3 | AVX2 | Yes | tensorflow_gpu-1.7.1-cp35-cp35m-win_amd64-gpu.whl |
| 1.7.1 | 3.5.3 | AVX2 | tensorflow-1.7.1-cp35-cp35m-win_amd64.whl | |
| 1.7.0 | 3.5.3 | AVX2 | tensorflow-1.7.0-cp35-cp35m-win_amd64.whl | |
| 1.7.0 | 3.5.3 | AVX2 | Yes | tensorflow_gpu-1.7.0-cp35-cp35m-win_amd64-avx2-cuda9.whl |
| 1.6.0 | 3.5.3 | AVX2 | tensorflow-1.6.0-cp35-cp35m-win_amd64-avx2.whl | |
| 1.5.1 | 3.5.3 | AVX2 | tensorflow-1.5.1-cp35-cp35m-win_amd64-avx2.whl | |
| 1.4.0 | 3.5.3 | AVX2 | tensorflow-1.4.0-cp35-cp35m-win_amd64.whl | |
| 1.4.0 | 3.5.3 | AVX | tensorflow-1.4.0-cp35-cp35m-win_amd64-avx.whl |
Build Environment
| Environment | Version |
|---|---|
| OS | Windows 10 x64 |
| Visual Sutdio | 2017 Community Edition for CPU only builds & 2015 Community Edition for CUDA enabled builds |
| CMAKE | 3.9.0 |
| SWIG | swigwin-3.0.12 |
| Python | 3.5.3 |
Followed official guide at https://github.com/tensorflow/tensorflow/blob/master/tensorflow/contrib/cmake/README.md
Build Commands
CMAKE Configuration Command for AVX/AVX2 non CUDA Builds using VS 2017
cmake.exe .. ^
-A x64 ^
-G "Visual Studio 15 2017" ^
-DSWIG_EXECUTABLE="c:/swigwin-3.0.12/swig.exe" ^
-DPYTHON_EXECUTABLE="c:/py/35/python.exe" ^
-DCMAKE_BUILD_TYPE=Release ^
-DPYTHON_LIBRARIES="c:/py/35/libs/python35.lib" ^
-Dtensorflow_BUILD_PYTHON_TESTS=OFF ^
-Dtensorflow_BUILD_CC_TESTS=OFF ^
-Dtensorflow_TF_NIGHTLY=OFF ^
-Dtensorflow_WIN_CPU_SIMD_OPTIONS=/arch:AVX2
Or when using CUDA, use the following command to build using toolchain 140 with VS 2015
cmake.exe .. ^
-A x64 ^
-G "Visual Studio 14 2015" -T v140 ^
-DSWIG_EXECUTABLE="C:/tools/swigwin-3.0.12/swig.exe" ^
-DPYTHON_EXECUTABLE="C:/Program Files/Python35/python.exe" ^
-DCMAKE_BUILD_TYPE=Release ^
-DPYTHON_LIBRARIES="C:/Program Files/Python35/libs/python35.lib" ^
-Dtensorflow_BUILD_PYTHON_TESTS=OFF ^
-Dtensorflow_BUILD_CC_TESTS=OFF ^
-Dtensorflow_TF_NIGHTLY=OFF ^
-Dtensorflow_WIN_CPU_SIMD_OPTIONS=/arch:AVX2 ^
-Dtensorflow_ENABLE_GPU=ON ^
-DCUDNN_HOME="C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v9.0"
Build with following command using "Developer Command Prompt for VS 2017" or "Developer Command Prompt for VS 2015"
msbuild.exe /p:Configuration=Release /maxcpucount:1 /verbosity:minimal tf_python_build_pip_package.vcxproj
Make sure 64bit build tools are used, rename "Hostx64" to "Hostx86" at following default install location for VS 2017
C:\Program Files (x86)\Microsoft Visual Studio\2017\Community\VC\Tools\MSVC\14.11.25503\bin
and for VS 2015 rename "amd64" to "x86_amd64" at the following default install location
C:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\bin