Win10环境Tensorflow-GPU13.1/JupyterNotebook的安装

 参考 ：

Anaconda Tensorflow GPU 版本的安装问题

https://blog.csdn.net/u010977034/article/details/62038698

Windows10环境下安装Anaconda和tensorflow-gpu，然后在jupyter notebook上使用

https://blog.csdn.net/qq_37392244/article/details/85003425

利用conda安装指定版本的tensorflow

conda install --channel https://conda.anaconda.org/anaconda tensorflow=1.6.0

Tensorflow-gpu搭建CUDA 10.0与cuDNN等版本问题

https://blog.csdn.net/weixin_42718092/article/details/85001140

---

查询Tensorflow与CUDA/CuDNN的版本关系

查询官网说明：

https://tensorflow.google.cn/install/source_windows

 查询CUDA要求的GUP Driver Version

 查询官网说明：

确定并安装需要的各个版本

python-3.6

tensorflow-gpu-1.13

cuda-10.0

cudnn-7.5

CUDA下载地址： https://developer.nvidia.com/cuda-toolkit-archive

CuDNN下载地址：https://developer.nvidia.com/rdp/cudnn-download

（下载CuDNN需要注册账号，注册过程比较简单）

说明：本次安装的tensorflow需要与Jupyter notebook集成，所以此处只安装CUDA和CuDNN（安装前确保GPU驱动已经安装且版本一致）。

CUDA是NVIDIA推出的运算平台，CuDNN是专门针对Deep Learning框架设计的一套GPU计算加速方案。在安装之前要查询下最新TensorFLow发行版支持到了哪个版本。另外，也要确认CUDA版本是否支持自己的显卡。

安装好ＣＵＤＡ后，再将下载好的ｃｕｄｎｎ文件解压缩，得到一个文件夹：

将里面的三个文件夹复制到C:Program FilesNVIDIA GPU Computing ToolkitCUDAv10.0中

---

Anaconda的下载和安装

安装Anaconda（一个开源的Python发行版本）的最新版本。Anaconda包含了conda、Python等180多个科学包及其依赖项，功能强大。
Anaconda下载地址：https://www.anaconda.com/download/

本次安装 的是Anaconda3-2018.12-Windows-x86_64，以下截图来自上述博客，主要用于说明有关的设置：

 创建tensorflow环境

 1、打开Anaconda Prompt

 

2、建立一个conda环境，命名为tensorflow_gpu

执行：conda create -n tensorflow_gpu python=3.6

3.进入到tensorflow_gpu环境中，在这个环境下面完成之后tensorflow-gpu的安装。

进入创建好的环境中，执行命令：activate tensorflow_gpu

安装tensorflow-gpu，执行命令：conda install --channel https://conda.anaconda.org/anaconda tensorflow-gpu

（该命令的查询方式，见下附【在Anaconda如何确定要安装的ｔｆ－ｇｐｕ版本】）

查看tensorlfow的版本：

 为了后面安装的tensorflow-gpu可以在jupyter notebook中使用 ，需要安装这两个包，用conda命令安装：

1. conda install ipython
2. conda install jupyter

执行命令：ipython kernelspec install-self --user

看到类似这个结果：Installed kernelspec python3 in C:UsersXXXJupyterkernelspython3

然后就可以使用jupyter notebook了，使用的时候先用activate tensorflow_gpu命令打开建立的环境，然后再用jupyter notebook命令打开jupyter notebook就可以了。

４、jupyter notebook的使用

 打开终端Anaconda Prompt，然后输入activate tensorflow_gpu，进入创建好的环境中，再用jupyter notebook命令打开jupyter notebook即可。

5、修改notebook文件的存放目录

 执行命令： jupyter notebook --no-browser --ip=0.0.0.0 --allow-root --NotebookApp.token= --notebook-dir='C:\Users\dell\MyJupyterCode'

在Anaconda如何确定要安装的ｔｆ－ｇｐｕ版本

参考：https://blog.csdn.net/u010977034/article/details/62038698

1、打开终端Anaconda Prompt，执行命令：anaconda search -t conda tensorflow

此指令会查询conda环境中有哪些tensorflow安装包，通过查看版本，选择最高的版本安装。如下图

 2、此处查询到的最高版本是 anaconda/tensorflow-gpu  1.13.1，则执行命令：anaconda show  anaconda/tensorflow-gpu 

这个命令令会告诉你怎么安装这个包

 

3、根据提示，在终端执行：

conda install --channel https://conda.anaconda.org/anaconda tensorflow-gpu

4、在第1步中可以看到，anaconda/tensorflow-gpu 带着版本号 1.13.1。因此，如果有多个版本，而你想安装一个指定的版本，则应该在命令后面加入版本信息，如：

conda install --channel https://conda.anaconda.org/anaconda tensorflow-gpu=1.13.1

如何删除已经建立 的conda环境

执行命令：

1. deactivate tensorflow_gpu
2. conda remove -n tensorflow --all
3. conda info --envs

    

    

---

 遇到的问题

 第一次安装后，在jupyter book中运行import tensorflow as tf时，报错。

重要的报错信息是：ImportError: DLL load failed: 找不到指定的模块。ModuleNotFoundError: No module named '_pywrap_tensorflow_internal'

解决方案：

　　Anaconda中安装 的tensorflow-gpu版本太低，与CUDA/Cudnn不匹配，重新按上述方法选择13.1安装即可。

--------------------详细报错信息如下-------------------------------------------------------
ImportError Traceback (most recent call last)
~AppDataLocalcondacondaenvs ensorflow_gpulibsite-packages ensorflowpythonpywrap_tensorflow_internal.py in swig_import_helper()
17 try:
---> 18 return importlib.import_module(mname)
19 except ImportError:

~AppDataLocalcondacondaenvs ensorflow_gpulibimportlib\__init__.py in import_module(name, package)
125 level += 1
--> 126 return _bootstrap._gcd_import(name[level:], package, level)
127

~AppDataLocalcondacondaenvs ensorflow_gpulibimportlib\_bootstrap.py in _gcd_import(name, package, level)

~AppDataLocalcondacondaenvs ensorflow_gpulibimportlib\_bootstrap.py in _find_and_load(name, import_)

~AppDataLocalcondacondaenvs ensorflow_gpulibimportlib\_bootstrap.py in _find_and_load_unlocked(name, import_)

~AppDataLocalcondacondaenvs ensorflow_gpulibimportlib\_bootstrap.py in _load_unlocked(spec)

~AppDataLocalcondacondaenvs ensorflow_gpulibimportlib\_bootstrap.py in module_from_spec(spec)

~AppDataLocalcondacondaenvs ensorflow_gpulibimportlib\_bootstrap_external.py in create_module(self, spec)

~AppDataLocalcondacondaenvs ensorflow_gpulibimportlib\_bootstrap.py in _call_with_frames_removed(f, *args, **kwds)

ImportError: DLL load failed: 找不到指定的模块。

During handling of the above exception, another exception occurred:

ModuleNotFoundError Traceback (most recent call last)
~AppDataLocalcondacondaenvs ensorflow_gpulibsite-packages ensorflowpythonpywrap_tensorflow.py in <module>()
40 sys.setdlopenflags(_default_dlopen_flags | ctypes.RTLD_GLOBAL)
---> 41 from tensorflow.python.pywrap_tensorflow_internal import *
42 from tensorflow.python.pywrap_tensorflow_internal import __version__

~AppDataLocalcondacondaenvs ensorflow_gpulibsite-packages ensorflowpythonpywrap_tensorflow_internal.py in <module>()
20 return importlib.import_module('_pywrap_tensorflow_internal')
---> 21 _pywrap_tensorflow_internal = swig_import_helper()
22 del swig_import_helper

~AppDataLocalcondacondaenvs ensorflow_gpulibsite-packages ensorflowpythonpywrap_tensorflow_internal.py in swig_import_helper()
19 except ImportError:
---> 20 return importlib.import_module('_pywrap_tensorflow_internal')
21 _pywrap_tensorflow_internal = swig_import_helper()

~AppDataLocalcondacondaenvs ensorflow_gpulibimportlib\__init__.py in import_module(name, package)
125 level += 1
--> 126 return _bootstrap._gcd_import(name[level:], package, level)
127

ModuleNotFoundError: No module named '_pywrap_tensorflow_internal'

During handling of the above exception, another exception occurred:

ImportError Traceback (most recent call last)
<ipython-input-1-41389fad42b5> in <module>()
----> 1 import tensorflow as tf

~AppDataLocalcondacondaenvs ensorflow_gpulibsite-packages ensorflow\__init__.py in <module>()
22
23 # pylint: disable=wildcard-import
---> 24 from tensorflow.python import *
25 # pylint: enable=wildcard-import
26

~AppDataLocalcondacondaenvs ensorflow_gpulibsite-packages ensorflowpython\__init__.py in <module>()
49 import numpy as np
50
---> 51 from tensorflow.python import pywrap_tensorflow
52
53 # Protocol buffers

~AppDataLocalcondacondaenvs ensorflow_gpulibsite-packages ensorflowpythonpywrap_tensorflow.py in <module>()
50 for some common reasons and solutions. Include the entire stack trace
51 above this error message when asking for help.""" % traceback.format_exc()
---> 52 raise ImportError(msg)
53
54 # pylint: enable=wildcard-import,g-import-not-at-top,unused-import,line-too-long

ImportError: Traceback (most recent call last):
File "C:UsersdellAppDataLocalcondacondaenvs ensorflow_gpulibsite-packages ensorflowpythonpywrap_tensorflow_internal.py", line 18, in swig_import_helper
return importlib.import_module(mname)
File "C:UsersdellAppDataLocalcondacondaenvs ensorflow_gpulibimportlib\__init__.py", line 126, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
File "<frozen importlib._bootstrap>", line 978, in _gcd_import
File "<frozen importlib._bootstrap>", line 961, in _find_and_load
File "<frozen importlib._bootstrap>", line 950, in _find_and_load_unlocked
File "<frozen importlib._bootstrap>", line 648, in _load_unlocked
File "<frozen importlib._bootstrap>", line 560, in module_from_spec
File "<frozen importlib._bootstrap_external>", line 922, in create_module
File "<frozen importlib._bootstrap>", line 205, in _call_with_frames_removed
ImportError: DLL load failed: 找不到指定的模块。

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "C:UsersdellAppDataLocalcondacondaenvs ensorflow_gpulibsite-packages ensorflowpythonpywrap_tensorflow.py", line 41, in <module>
from tensorflow.python.pywrap_tensorflow_internal import *
File "C:UsersdellAppDataLocalcondacondaenvs ensorflow_gpulibsite-packages ensorflowpythonpywrap_tensorflow_internal.py", line 21, in <module>
_pywrap_tensorflow_internal = swig_import_helper()
File "C:UsersdellAppDataLocalcondacondaenvs ensorflow_gpulibsite-packages ensorflowpythonpywrap_tensorflow_internal.py", line 20, in swig_import_helper
return importlib.import_module('_pywrap_tensorflow_internal')
File "C:UsersdellAppDataLocalcondacondaenvs ensorflow_gpulibimportlib\__init__.py", line 126, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
ModuleNotFoundError: No module named '_pywrap_tensorflow_internal'

Failed to load the native TensorFlow runtime.

See https://www.tensorflow.org/install/install_sources#common_installation_problems

for some common reasons and solutions. Include the entire stack trace
above this error message when asking for help.