TensorFlow导入报错：找不到libcusolver.so.8.0文件该如何解决？

我之前也碰到过完全一样的报错，当执行import tensorflow as tf时，系统找不到libcusolver.so.8.0这个共享库，直接导致TensorFlow原生运行时加载失败。先把你遇到的完整报错贴出来方便大家参考：

import tensorflow as tf

Traceback (most recent call last): File "/home/user/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow.py", line 41, in <module> from tensorflow.python.pywrap_tensorflow_internal import * File "/home/user/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 28, in <module> _pywrap_tensorflow_internal = swig_import_helper() File "/home/user/anaconda3/lib/python3.6/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 "/home/user/anaconda3/lib/python3.6/imp.py", line 243, in load_module return load_dynamic(name, filename, file) File "/home/user/anaconda3/lib/python3.6/imp.py", line 343, in load_dynamic return _load(spec) ImportError: libcusolver.so.8.0: cannot open shared object file: No such file or directory During handling of the above exception, another exception occurred: Traceback (most recent call last): File "<stdin>", line 1, in <module> File "/home/user/anaconda3/lib/python3.6/site-packages/tensorflow/__init__.py", line 24, in <module> from tensorflow.python import * File "/home/user/anaconda3/lib/python3.6/site-packages/tensorflow/python/__init__.py", line 49, in <module> from tensorflow.python import pywrap_tensorflow File "/home/user/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow.py", line 52, in <module> raise ImportError(msg) ImportError: Traceback (most recent call last): File "/home/user/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow.py", line 41, in <module> from tensorflow.python.pywrap_tensorflow_internal import * File "/home/user/anaconda3/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 28, in <module> _pywrap_tensorflow_internal = swig_import_helper() File "/home/user/anaconda3/lib/python3.6/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 "/home/user/anaconda3/lib/python3.6/imp.py", line 243, in load_module return load_dynamic(name, filename, file) File "/home/user/anaconda3/lib/python3.6/imp.py", line 343, in load_dynamic return _load(spec) ImportError: libcusolver.so.8.0: cannot open shared object file: No such file or directory Failed to load the native TensorFlow runtime.

核心原因 #

这个报错的本质是你的系统缺少TensorFlow依赖的CUDA 8.0组件，或者CUDA 8.0的库路径没有被系统正确识别到。libcusolver.so.8.0是NVIDIA CUDA 8.0提供的线性求解库，旧版本的TensorFlow（比如1.x系列）对CUDA版本有严格的对应要求。

解决方案 #

1. 确认CUDA 8.0是否已安装

先检查系统里有没有CUDA 8.0：

ls /usr/local/cuda-8.0/lib64/libcusolver.so.8.0

如果返回文件不存在，说明你需要安装CUDA 8.0。注意要匹配你TensorFlow版本对应的CUDA要求（旧TF版本大多对应CUDA 8.0/9.0）。

2. 配置环境变量（已安装CUDA但路径未识别）

如果已经安装了CUDA 8.0，但系统找不到库，需要把CUDA的lib路径添加到LD_LIBRARY_PATH：

• 临时生效（当前终端）：

  export LD_LIBRARY_PATH=/usr/local/cuda-8.0/lib64:$LD_LIBRARY_PATH
  

  执行完这个命令后，再尝试导入TensorFlow看看是否解决。
• 永久生效：
  如果你用的是bash，编辑~/.bashrc文件，添加上面的export命令，然后执行：

  source ~/.bashrc
  

  如果你用的是zsh，就编辑~/.zshrc，同样source生效。

3. 检查cuDNN版本是否匹配

除了CUDA，还要确保cuDNN的版本和CUDA 8.0兼容，比如cuDNN 5.1或6.0都是适配CUDA 8.0的版本，把cuDNN的库也放到CUDA的lib路径下，或者添加到LD_LIBRARY_PATH里。

内容的提问来源于stack exchange，提问作者ellibria