当我命令conda install tensorflow-gpu
时,Anaconda将在conda环境的packages目录中安装cudnn-7.6.4的新版本。
Tensorflow使用的是新版本的cudnn,这在运行我的代码时在下面导致cudnn错误。
tensorflow.python.framework.errors_impl.UnknownError: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize,so try looking to see if a warning log message was printed above. [Op:Conv2D]
如何将cudnn版本更改为7.4.1?
和 当我再次查看该错误时
Loaded runtime CuDNN library: 7.4.1 but source was compiled with: 7.6.0. CuDNN library major and minor version needs to match or have higher minor version in case of CuDNN 7.0 or later version.
If using a binary install,upgrade your CuDNN library.
If building from sources,make sure the library loaded at runtime is compatible with the version specified during compile configuration.
2019-11-14 14:54:56.745762: E tensorflow/stream_executor/cuda/cuda_dnn.cc:319] Loaded runtime CuDNN library: 7.4.1 but source was compiled with: 7.6.0. CuDNN library major and minor version needs to match or have higher minor version in case of CuDNN 7.0 or later version.
If using a binary install,make sure the library loaded at runtime is compatible with the version specified during compile configuration.
但是我输入以下命令
conda list cudnn
我有
# packages in environment at C:\Users\user\Anaconda3\envs\arcface-tf2:
#
# Name Version Build Channel
cudnn 7.6.4 cuda10.0_0
我的cudnn版本是什么?