Given 1: Use after installation pip Navigate tensorflow-gpu but inside the corresponding environments in which not find Anaconda inside Terminal;
Workaround: pip conda can be replaced, or directly in environments which installs the specified version can;
2 being given: cuda installed and cudnn must correspond to the version that is installed tensorflow; specific reference to the following:
cuDNN Download: https://developer.nvidia.com/rdp/cudnn- download , download cuDNN too much trouble, need to register an account after answering some questions before they can enter download CUDA Downloads: HTTPS: //developer.nvidia.com/cuda-toolkit-archive , download the corresponding version on demand
Error 3: CUDA association and cuDNN
(1) after extracting cuDNN compressed, the files are decompressed copy and paste the corresponding folder cuda
(2) after decompression cuDNN compressed, the files are decompressed copy and paste the corresponding folder cuda
TensorFlow GPU test environment is successfully installed
1) Anaconda Navigator to open the corresponding virtual environment Terminal, enter python, python into the environment
2) according to the following FIG enter command line displays information representing the computer graphics some success
