python - Tensorflow 导入错误

Question

我正在尝试安装支持 GPU 的 tensorflow。

我尝试了以下链接中的信息

https://www.tensorflow.org/install/install_windows

1. CUDA® 工具包 8.0
2. cuDNN v6.0
3. 具有 CUDA Compute Capability 3.0 的 GPU 卡 - GeForce 940MX

然后使用pip3 install --upgrade tensorflow-gpu安装tensorflow。

但是我在尝试导入 tensorflow 时遇到以下错误。

    Traceback (most recent call last):
      File "C:\Research\Python_installation\lib\site-packages\tensorflow\python\platform\self_check.py", line 75, in preload_check
        ctypes.WinDLL(build_info.cudart_dll_name)
      File "C:\Research\Python_installation\lib\ctypes\__init__.py", line 347, in __init__
        self._handle = _dlopen(self._name, mode)
    OSError: [WinError 126] The specified module could not be found

    During handling of the above exception, another exception occurred:

    Traceback (most recent call last):
      File "<pyshell#6>", line 1, in <module>
        import tensorflow as tf
      File "C:\Research\Python_installation\lib\site-packages\tensorflow\__init__.py", line 24, in <module>
        from tensorflow.python import *
      File "C:\Research\Python_installation\lib\site-packages\tensorflow\python\__init__.py", line 49, in <module>
        from tensorflow.python import pywrap_tensorflow
      File "C:\Research\Python_installation\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 30, in <module>
        self_check.preload_check()
      File "C:\Research\Python_installation\lib\site-packages\tensorflow\python\platform\self_check.py", line 82, in preload_check
        % (build_info.cudart_dll_name, build_info.cuda_version_number))
    ImportError: Could not find 'cudart64_90.dll'. TensorFlow requires that this DLL be installed in a directory that is named in your %PATH% environment variable. Download and install CUDA 9.0 from this URL: https://developer.nvidia.com/cuda-toolkit

有人可以建议我如何使用它吗。

Answer 1

我有一个类似的问题，必须非常小心 CUDA 的版本和 CuDNN 的版本。我遇到了您遇到的确切错误，并通过我在此处记录的内容修复了它:http://www.laurencemoroney.com/installing-tensorflow-with-gpu-on-windows-10/

试试吧! :)

(我发现最常见的失败是您下载了最新 CUDA，而不是匹配 CUDA。现在 CUDA 是 9.1，但 TF 需要9.0 -- 你的错误是 cudart90.dll) -- 所以找到 9.0 驱动程序，下载并安装它们。然后运行 ​​TensorFlow。它可能会在 CuDNN 驱动程序上失败，这很好，因为您知道 CUDA 是正确的。然后下载正确的 CuDNN 驱动程序(与错误中的版本号匹配)并重试。)