Windows 10安装TensorFlow-gpu1.4 及CUDA8.0,cuDNN6.0，搞定了
先说下我的最终版本：
Python 3.5.3；
tensorflow-gpu==1.4.0 ；
CUDA Toolkit 8.0 GA1 (Sept 2016), Online Documentation；
cuDNN v6.0（Download cuDNN v6.0 (April 27, 2017), for CUDA 8.0）
注意CUDA的版本是由TensorFlow决定的。具体如何对应，需要去TensorFlow官网看：https://www.tensorflow.org/install/install_windows
https://www.tensorflow.org/install/install_windows#requirements_to_run_tensorflow_with_gpu_support（很抱歉，大陆打不开2018.9.10），还有个方法是，下载喜欢的TF版本，让系统报错提示你安装合适的CUDA版本
安装TensorFlow：pip install tensorflow-gpu==1.4.0
下载CUDA8.0 :https://developer.nvidia.com/cuda-toolkit-archive
CUDA Toolkit 8.0 GA1 (Sept 2016), Online Documentation
安装方法：https://docs.nvidia.com/cuda/archive/8.0/cuda-installation-guide-microsoft-windows/index.html
路径C:\Users\guang\AppData\Local\Temp\CUDA
验证CUDA：
控制栏：
C:\Users\guang>nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2016 NVIDIA Corporation
Built on Sat_Sep__3_19:05:48_CDT_2016
Cuda compilation tools, release 8.0, V8.0.44
下载cuDNN：https://developer.nvidia.com/rdp/cudnn-archive 需要注册账号
Download cuDNN v7.0.5 (Dec 5, 2017), for CUDA 8.0（注意切换到6.0，要不然跟着我找麻烦）
安装cuDNN步骤，比较麻烦，注意看，主要就是把cuDNN三个文件复制到CUDA对应文件夹下就行
安装方法：https://docs.nvidia.com/deeplearning/sdk/cudnn-install/index.html#download-windows
第五步我没有做，因为没有对应的VS软件
Include cudnn.lib in your Visual Studio project.
Open the Visual Studio project and right-click on the project name.
Click Linker > Input > Additional Dependencies.
Add cudnn.lib and click OK.
运行代码测试效果：
import tensorflow as tf
a = tf.random_normal((100, 100))
b = tf.random_normal((100, 500))
c = tf.matmul(a, b)
sess = tf.InteractiveSession()
报错1：
OSError: [WinError 126] 找不到指定的模块。
ImportError: Could not find 'msvcp140.dll'. TensorFlow requires that this DLL be installed in a directory that is named in your %PATH% environment variable. You may install this DLL by downloading Visual C++ 2015 Redistributable Update 3 from this URL: https://www.microsoft.com/en-us/download/details.aspx?id=53587
解决方案：
下载软件：
https://www.microsoft.com/en-us/download/details.aspx?id=53587
64位电脑下载软件vc_redist.x64.exe，双击安装
vc_redist.x64.exe而不是
vc_redist.x86.exe
x86和X64是什么意思
x64 是指CPU是64位版本的。
x86 是指CPU是32位版本的。
如果你的CPU是64位的。可以安装64位的，也可以安装32位的，反过来只能安装32位的。
报错2：
ImportError: Could not find 'cudart64_80.dll'. TensorFlow requires that this DLL be installed in a directory that is named in your %PATH% environment variable. Download and install CUDA 8.0 from this URL: https://developer.nvidia.com/cuda-toolkit
下载软件2：
https://developer.nvidia.com/cuda-toolkit
这一步是下载工具箱cuda-toolkit（其实就是CUDA），下载后是cuda_9.2.148_win10.exe格式，我明明装的是CUDA8.0，符合这里的要求。Download and install CUDA 8.0 from this URL奇怪了
后来我通过软件search everthing找到了文件cudart64_80.dll，然后把他添加到系统路径，命名为：CUDA_PATH_cudart64_80,路径为：C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v8.0\bin
重启电脑解决了。
再次运行，再次报错：
ImportError: Could not find 'cudnn64_6.dll'. TensorFlow requires that this DLL be installed in a directory that is named in your %PATH% environment variable. Note that installing cuDNN is a separate step from installing CUDA, and this DLL is often found in a different directory from the CUDA DLLs. You may install the necessary DLL by downloading cuDNN 6 from this URL: https://developer.nvidia.com/cudnn
找不到这个文件：cudnn64_6.dll，我搜索了一下，确实没有找到，他提示我安装cuDNN的cuDNN 6，我突然想起来，自己装了cuDNN 7（Download cuDNN v7.0.5 (Dec 5, 2017), for CUDA 8.0）
我无语了，官网自己说7.0也是配套CUDA8的嘛。没办法，按照提示来吧，下载Download cuDNN v6.0 (April 27, 2017), for CUDA 8.0，三个文件复制到CUDA对应目录里，再次运行，显示如下内容，成功了。说明我的GPU是GeForce GTX 1050 Ti，6个运算力，满足3个就够了，频率：1.475，内存：4G，
欢迎关注《湾区人工智能》公众号
2018-09-11 09:16:26.582331: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\35\tensorflow\core\platform\cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX AVX2
2018-09-11 09:16:27.016004: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\35\tensorflow\core\common_runtime\gpu\gpu_device.cc:1030] Found device 0 with properties:
name: GeForce GTX 1050 Ti major: 6 minor: 1 memoryClockRate(GHz): 1.4175
pciBusID: 0000:01:00.0
totalMemory: 4.00GiB freeMemory: 3.30GiB
2018-09-11 09:16:27.016603: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\35\tensorflow\core\common_runtime\gpu\gpu_device.cc:1120] Creating TensorFlow device (/device:GPU:0) -> (device: 0, name: GeForce GTX 1050 Ti, pci bus id: 0000:01:00.0, compute capability: 6.1)
[Finished in 2.4s]
---------------------
作者：湾区Python
来源：CSDN
原文：https://blog.csdn.net/BTUJACK/article/details/82843200
版权声明：本文为博主原创文章，转载请附上博文链接！