TensorFlow、cuda、cudnn的版本到底该怎么选择

Question

首先我的电脑显卡是1650ti，适合的cuda版本应该是11.3，安装的cudnn版本是8.2.1。现在想要运行一段代码，需要的环境是tensorflow-gpu==1.13.1，keras==2.1.5。使用anaconda配置了相应的环境

python的版本是3.6.13

import tensorflow之后报错

Traceback (most recent call last):
  File "E:\anaconda3\envs\vgg\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 58, in <module>
    from tensorflow.python.pywrap_tensorflow_internal import *
  File "E:\anaconda3\envs\vgg\lib\site-packages\tensorflow\python\pywrap_tensorflow_internal.py", line 28, in <module>
    _pywrap_tensorflow_internal = swig_import_helper()
  File "E:\anaconda3\envs\vgg\lib\site-packages\tensorflow\python\pywrap_tensorflow_internal.py", line 24, in swig_import_helper
    _mod = imp.load_module('_pywrap_tensorflow_internal', fp, pathname, description)
  File "E:\anaconda3\envs\vgg\lib\imp.py", line 243, in load_module
    return load_dynamic(name, filename, file)
  File "E:\anaconda3\envs\vgg\lib\imp.py", line 343, in load_dynamic
    return _load(spec)
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 "<stdin>", line 1, in <module>
  File "E:\anaconda3\envs\vgg\lib\site-packages\tensorflow\__init__.py", line 24, in <module>
    from tensorflow.python import pywrap_tensorflow  # pylint: disable=unused-import
  File "E:\anaconda3\envs\vgg\lib\site-packages\tensorflow\python\__init__.py", line 49, in <module>
    from tensorflow.python import pywrap_tensorflow
  File "E:\anaconda3\envs\vgg\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 74, in <module>
    raise ImportError(msg)
ImportError: Traceback (most recent call last):
  File "E:\anaconda3\envs\vgg\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 58, in <module>
    from tensorflow.python.pywrap_tensorflow_internal import *
  File "E:\anaconda3\envs\vgg\lib\site-packages\tensorflow\python\pywrap_tensorflow_internal.py", line 28, in <module>
    _pywrap_tensorflow_internal = swig_import_helper()
  File "E:\anaconda3\envs\vgg\lib\site-packages\tensorflow\python\pywrap_tensorflow_internal.py", line 24, in swig_import_helper
    _mod = imp.load_module('_pywrap_tensorflow_internal', fp, pathname, description)
  File "E:\anaconda3\envs\vgg\lib\imp.py", line 243, in load_module
    return load_dynamic(name, filename, file)
  File "E:\anaconda3\envs\vgg\lib\imp.py", line 343, in load_dynamic
    return _load(spec)
ImportError: DLL load failed: 找不到指定的模块。


Failed to load the native TensorFlow runtime.

See https://www.tensorflow.org/install/errors

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

CSDN了一下，一个原因是缺少库MSVCP140.dll

经检查并不是
还有可能是缺少用于Visual Studio 2015的Visual C ++可再发行组件而引起的

经检查也不是
还有一个原因是TensorFlow和cuda的版本不匹配
搜索了一下TensorFlow_gpu1.13.1对应的cuda版本好像是10.1，但是cuda10.1和我的显卡不匹配，所以这个问题应该怎么解决呢

Answer 1

Build from source on Windows  |  TensorFlow https://tensorflow.google.cn/install/source_windows?hl=en#cpu

去链接里面看，cuda和tf版本要对应的上才行
不匹配的话有两种，一种是保持cuda，然后换tf版本。另外一种就是安装两个版本的cuda，通过环境变量来控制环境重的cuda版本，实现切换。另外一种就是在conda的虚拟环境中安装另外的cuda版本（不推荐）