1.setuptools 版本过旧需要更新

ERROR: tensorboard 2.0.0 has requirement setuptools>=41.0.0, but you'll have set
uptools 36.5.0.post20170921 which is incompatible.

  解决方式:

pip install --upgrade setuptools

2.文件wrapt不确定归属

ERROR: Cannot uninstall 'wrapt'. It is a distutils installed project and thus we
 cannot accurately determine which files belong to it which would lead to only a
 partial uninstall.

  解决方式:

pip install -U --ignore-installed wrapt enum34 simplejson netaddr

3.驱动版本不能支撑cuda运行版本

tensorflow.python.framework.errors_impl.InternalError: cudaGetDevice() failed. S
tatus: CUDA driver version is insufficient for CUDA runtime version

  解决方式: 更新即可

https://zhidao.baidu.com/question/1951584892469977108.html

  tensorflow1.8-gpu安装注意

cuda9.0+cudnn7.0配套安装

[补充:我的是Geforce 940M,重新安装了显卡驱动397.93-notebook-win10-64bit-international-whql.exe(可能不需要) 下载地址:https://www.geforce.cn/drivers(驱动要与显卡型号匹配)]

cuda9.0官网下载:https://developer.nvidia.com/cuda-toolkit-archive

cudnn7.0官网下载:https://developer.nvidia.com/rdp/cudnn-archive

5.命令行输入:nvcc -V //查看cuda是否安装好

说明:将cudnn7.0文件夹bin、include、lib中的文件分别拷贝至cuda9.0安装目录下的对应文件夹中,如我的C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v9.0里的bin、include、lib文件夹内