ubuntu16.4+anaconda+cuda9.0+cudnn7.0.5+tensorflow-gpu1.9.0+keras安装参考

从国内的开源镜像站下载，如清华大学开源软件镜像站，这里我选择下载Anaconda3-5.0.0-Linux-x86_64.sh。(https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/)

https://blog.csdn.net/huangjuegeek/article/details/73556763   

https://blog.csdn.net/qq_20492405/article/details/79034430  Ubuntu16.04+nvidia显卡驱动+cuda安装

Ubuntu 16.04 上安装 CUDA 9.0 详细教程

https://blog.csdn.net/QLULIBIN/article/details/78714596#commentBox

 

安装cudnn7.0.5

https://www.cnblogs.com/pprp/p/9540523.html#cuda-9.0%E5%AE%89%E8%A3%85

tensorflow-gpu

python版本动态配置

https://blog.csdn.net/davidhopper/article/details/81206673

https://blog.csdn.net/weixin_39345190/article/details/82595898

https://blog.csdn.net/qq_34289157/article/details/88071116在 Anaconda下解决国内安装tensorflow等下载慢和中断，出错，异常问题的一点思路

https://blog.csdn.net/AllenIrving/article/details/80406709  pip换源加速

Windows下，pip安装时ReadTimeoutError解决办法

https://blog.csdn.net/abcabc77777/article/details/53456453

numpy版本太高会和tensorflow1.9.0不符，pip uninstall numpy卸载再安装>=1.13.3,安装1.14.0

警告：Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA

https://www.cnblogs.com/Ph-one/p/9033733.html

 

深度学习（TensorFlow）环境搭建：（二）Ubuntu16.04+CUDA9.0+cuDNN7+Anaconda5.1+Python3.6+TensorFlow1.7  

https://blog.csdn.net/fnhc462354756/article/details/80041021

WIN10安装CUDA9+CUDNN+TENSORFLOW1.9.0+KERAS

https://www.jianshu.com/p/c228bd790775    

(有keras测试用例)

prcharm 安装 https://blog.csdn.net/zhuanshu666/article/details/73554885