Build from source on Windows

Build a TensorFlow pip package from the source and install it on Windows.

Setup for Windows

Install the following build tools to configure your Windows development environment.

Install Python and the TensorFlow package dependencies

Install a Python 3.9 64-bit release for Windows. Select pip as an optional feature and add it to your %PATH% environmental variable.

Install the TensorFlow pip package dependencies:

pip3 install -U pip
pip3 install -U six numpy wheel packaging
pip3 install -U keras_preprocessing --no-deps

The dependencies are listed in the setup.py file under REQUIRED_PACKAGES.

Install Bazel

Install Bazel, the build tool used to compile TensorFlow. For Bazel version, see the tested build configurations for Windows. Configure Bazel to build C .

Add the location of the Bazel executable to your %PATH% environment variable.

Install MSYS2

Install MSYS2 for the bin tools needed to build TensorFlow. If MSYS2 is installed to C:\msys64, add C:\msys64\usr\bin to your %PATH% environment variable. Then, using cmd.exe, run:

pacman -Syu (requires a console restart)
pacman -S git patch unzip
pacman -S git patch unzip rsync

Install Visual C Build Tools 2022

Install the Visual C build tools 2022. This comes with Visual Studio Community 2022 but can be installed separately:

1. Go to the Visual Studio downloads,
2. Select Tools for Visual Studio or Other Tools, Framework and Redistributables,
3. Download and install:
   • Build Tools for Visual Studio 2022
   • Microsoft Visual C Redistributables for Visual Studio 2022

Install LLVM

1. Go to the LLVM downloads,
2. Download and install Windows-compatible LLVM in C:/Program Files/LLVM e.g., LLVM-17.0.6-win64.exe

Install GPU support (optional)

See the Windows GPU support guide to install the drivers and additional software required to run TensorFlow on a GPU.

Download the TensorFlow source code

Use Git to clone the TensorFlow repository (git is installed with MSYS2):

git clone https://github.com/tensorflow/tensorflow.git
cd tensorflow

The repo defaults to the master development branch. You can also check out a release branch to build:

git checkout branch_name  # r1.9, r1.10, etc.

Optional: Environmental Variable Set Up

Run the following commands before running the build command to avoid issues with package creation: (If the below commands were set up while installing the packages, please ignore them). Run set to check if all the paths were set correctly, run echo %Environmental Variable% e.g., echo �ZEL_VC% to check the path set up for a specific Environmental Variable

Python path set up issue tensorflow:issue#59943,tensorflow:issue#9436,tensorflow:issue#60083

set PATH=path/to/python;%PATH% # [e.g. (C:/Python311)]
set PATH=path/to/python/Scripts;%PATH% # [e.g. (C:/Python311/Scripts)] 
set PYTHON_BIN_PATH=path/to/python_virtualenv/Scripts/python.exe 
set PYTHON_LIB_PATH=path/to/python virtualenv/lib/site-packages 
set PYTHON_DIRECTORY=path/to/python_virtualenv/Scripts 

Bazel/MSVC/CLANG path set up issue tensorflow:issue#54578

set BAZEL_SH=C:/msys64/usr/bin/bash.exe 
set BAZEL_VS=C:/Program Files/Microsoft Visual Studio/2022/BuildTools 
set BAZEL_VC=C:/Program Files/Microsoft Visual Studio/2022/BuildTools/VC 
set Bazel_LLVM=C:/Program Files/LLVM (explicitly tell Bazel where LLVM is installed by BAZEL_LLVM, needed while using CLANG)
set PATH=C:/Program Files/LLVM/bin;%PATH% (Optional, needed while using CLANG as Compiler)

Optional: Configure the build

TensorFlow builds are configured by the .bazelrc file in the repository's root directory. The ./configure or ./configure.py scripts can be used to adjust common settings.

If you need to change the configuration, run the ./configure script from the repository's root directory.

python ./configure.py

This script prompts you for the location of TensorFlow dependencies and asks for additional build configuration options (compiler flags, for example). The following shows a sample run of python ./configure.py (your session may differ):

python ./configure.py
You have bazel 6.5.0 installed.
Please specify the location of python. [Default is C:\Python311\python.exe]:

Found possible Python library paths:
C:\Python311\lib\site-packages
Please input the desired Python library path to use.  Default is [C:\Python311\lib\site-packages]

Do you wish to build TensorFlow with ROCm support? [y/N]:
No ROCm support will be enabled for TensorFlow.

Warning: Cannot build with CUDA support on Windows.
Starting in TF 2.11, CUDA build is not supported for Windows. To use TensorFlow GPU on Windows, you will need to build/install TensorFlow in WSL2.

Do you want to use Clang to build TensorFlow? [Y/n]:
Add "--config=win_clang" to compile TensorFlow with CLANG.

Please specify the path to clang executable. [Default is C:\Program Files\LLVM\bin\clang.EXE]:

You have Clang 17.0.6 installed.

Please specify optimization flags to use during compilation when bazel option "--config=opt" is specified [Default is /arch:AVX]:

Would you like to override eigen strong inline for some C   compilation to reduce the compilation time? [Y/n]:
Eigen strong inline overridden.

Would you like to interactively configure ./WORKSPACE for Android builds? [y/N]:
Not configuring the WORKSPACE for Android builds.

Preconfigured Bazel build configs. You can use any of the below by adding "--config=<>" to your build command. See .bazelrc for more details.
        --config=mkl            # Build with MKL support.
        --config=mkl_aarch64    # Build with oneDNN and Compute Library for the Arm Architecture (ACL).
        --config=monolithic     # Config for mostly static monolithic build.
        --config=numa           # Build with NUMA support.
        --config=dynamic_kernels        # (Experimental) Build kernels into separate shared objects.
        --config=v1             # Build with TensorFlow 1 API instead of TF 2 API.
Preconfigured Bazel build configs to DISABLE default on features:
        --config=nogcp          # Disable GCP support.
        --config=nonccl         # Disable NVIDIA NCCL support.

Build and install the pip package

The pip package is built in two steps. A bazel build command creates a "package-builder" program. You then run the package-builder to create the package.

Build the package-builder

tensorflow:master repo has been updated to build 2.x by default. Install Bazel and use bazel build to create the TensorFlow package-builder.

bazel build //tensorflow/tools/pip_package:wheel

CPU-only

Use bazel to make the TensorFlow package builder with CPU-only support:

Build with MSVC

bazel build --config=opt --repo_env=TF_PYTHON_VERSION=3.11 //tensorflow/tools/pip_package:wheel --repo_env=WHEEL_NAME=tensorflow_cpu

Build with CLANG

Use --config=win_clang to build TenorFlow with the CLANG Compiler:

bazel build --config=win_clang --repo_env=TF_PYTHON_VERSION=3.11 //tensorflow/tools/pip_package:wheel --repo_env=WHEEL_NAME=tensorflow_cpu

GPU support

To make the TensorFlow package builder with GPU support:

bazel build --config=opt --config=cuda --define=no_tensorflow_py_deps=true //tensorflow/tools/pip_package:build_pip_package

Commands to clean the bazel cache to resolve errors due to invalid or outdated cached data, bazel clean with --expunge flag removes files permanently

bazel clean 
bazel clean --expunge  

Bazel build options

Use this option when building to avoid issues with package creation: tensorflow:issue#22390

--define=no_tensorflow_py_deps=true

See the Bazel command-line reference for build options.

Building TensorFlow from source can use a lot of RAM. If your system is memory-constrained, limit Bazel's RAM usage with: --local_ram_resources=2048.

If building with GPU support, add --copt=-nvcc_options=disable-warnings to suppress nvcc warning messages.

Build the package

To build a pip package, you need to specify the --repo_env=WHEEL_NAME flag. Depending on the provided name, the package will be created. For example:

To build tensorflow CPU package:

bazel build //tensorflow/tools/pip_package:wheel --repo_env=WHEEL_NAME=tensorflow_cpu

To build nightly package, set tf_nightly instead of tensorflow, e.g. to build CPU nightly package:

bazel build //tensorflow/tools/pip_package:wheel --repo_env=WHEEL_NAME=tf_nightly_cpu

As a result, generated wheel will be located in

bazel-bin/tensorflow/tools/pip_package/wheel_house/

Install the package

The filename of the generated .whl file depends on the TensorFlow version and your platform. Use pip install to install the package, for example:

pip install bazel-bin/tensorflow/tools/pip_package/wheel_house/tensorflow-version-tags.whl

Build using the MSYS shell

TensorFlow can also be built using the MSYS shell. Make the changes listed below, then follow the previous instructions for the Windows native command line (cmd.exe).

Disable MSYS path conversion

MSYS automatically converts arguments that look like Unix paths to Windows paths, and this doesn't work with bazel. (The label //path/to:bin is considered a Unix absolute path since it starts with a slash.)

export MSYS_NO_PATHCONV=1
export MSYS2_ARG_CONV_EXCL="*"

Set your PATH

Add the Bazel and Python installation directories to your $PATH environmental variable. If Bazel is installed to C:\tools\bazel.exe, and Python to C:\Python\python.exe, set your PATH with:

# Use Unix-style with ':' as separator
export PATH="/c/tools:$PATH"
export PATH="/c/path/to/Python:$PATH"

For GPU support, add the CUDA and cuDNN bin directories to your $PATH:

export PATH="/c/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.0/bin:$PATH"
export PATH="/c/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.0/extras/CUPTI/libx64:$PATH"
export PATH="/c/tools/cuda/bin:$PATH"

Tested build configurations

CPU

GPU