编译好的包:https://github.com/fo40225/tensorflow-windows-wheel
官方编译教程:https://tensorflow.google.cn/install/source_windows?hl=zh-cn
参考教程:http://www.leoxiaofei.com/build-tensorflow-windows.html
- 安装:visual studio 2019、Python、Bazel、MSYS2
- tensorflow源码:https://github.com/tensorflow/tensorflow.git
- 运行:GPU版本
在开始菜单里找到并打开vs2017编译命令行:"x64 Native Tools Command Prompt for VS 2017"
cd 到 tensorflow 源码目录
python ./configure.py
输入python configure.py,一般都用默认值,GPU版中"Do you wish to build TensorFlow with CUDA support? [y/N]:"回答y
TensorFlow 2.x
bazel build //tensorflow/tools/pip_package:build_pip_package
TensorFlow 1.x
bazel build --config=v1 //tensorflow/tools/pip_package:build_pip_package
仅支持 CPU
bazel build --config=opt //tensorflow/tools/pip_package:build_pip_package
GPU 支持
bazel build --config=opt --config=cuda --define=no_tensorflow_py_deps=true //tensorflow/tools/pip_package:build_pip_package
编译推荐用下面的 生成dll和lib文件(注意内存虽然写了2048,16G内存还是不够报错,最好32G)(推荐用编译好的包:https://github.com/fo40225/tensorflow-windows-wheel)
//生成whl包
bazel build --config=opt --config=v1 --config=cuda --copt=-nvcc_options=disable-warnings --define=no_tensorflow_py_deps=true --local_ram_resources=2048 //tensorflow/tools/pip_package:build_pip_package
//生成os文件
bazel build --config=opt --config=v1 --config=cuda --copt=-nvcc_options=disable-warnings --define=no_tensorflow_py_deps=true --local_ram_resources=2048 //tensorflow:libtensorflow_cc.so
//生成dll和lib文件
bazel build --config=opt --config=v1 --config=cuda --copt=-nvcc_options=disable-warnings --define=no_tensorflow_py_deps=true --local_ram_resources=2048 //tensorflow:tensorflow_cc.dll
- 等着。。。。