Could not load dynamic library ‘libcudart.so.10.0‘

2022-04-29 03:35:16.853021: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcudart.so.10.0'; dlerror: libcudart.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-10.1/lib64
2022-04-29 03:35:16.853249: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcublas.so.10.0'; dlerror: libcublas.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-10.1/lib64
2022-04-29 03:35:16.853461: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcufft.so.10.0'; dlerror: libcufft.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-10.1/lib64
2022-04-29 03:35:16.853664: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcurand.so.10.0'; dlerror: libcurand.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-10.1/lib64
2022-04-29 03:35:16.853869: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcusolver.so.10.0'; dlerror: libcusolver.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-10.1/lib64
2022-04-29 03:35:16.854067: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcusparse.so.10.0'; dlerror: libcusparse.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/cuda-10.1/lib64

出现问题的原因

最后在网上搜索发现问题出现的根本原因在于CUDA版本（10.1）和TensorFlow版本（1.14.0）不一致造成的。因为最新的CUDA版本已经更新到10.1+，但是TensorFlow最新只支持到10.0，所以才会出现各种找不到10.0的库。为什么耗费了这么久？因为CUDA很狡猾，在用nvcc -V命令查看时，给出的版本号是10.0，但用watch -n 1 nvidia -smi查询时，右上角显示的却是10.1。这里推荐用下面这个命令查询：

(Shugang3DGCN) usr@ubuntu16:~$ conda list

在其中找到cudatoolkit这一项，这个版本号是比较准确的。

解决办法

我这显示的cudatoolkit版本是10.1，因为需要10.0的版本，可以用如下命令直接覆盖安装

(Shugang3DGCN) usr@ubuntu16:~$ conda install cudatoolkit=10.0

最后，再用conda list查询是否覆盖成功：

在改成10.0之后，代码应该就不会再出问题了。

link