tensorflow Installation

问题1

>>> import tensorflow as tf
2023-05-31 10:50:34.908411: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2023-05-31 10:50:35.606378: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer.so.7'; dlerror: libnvinfer.so.7: cannot open shared object file: No such file or directory
2023-05-31 10:50:35.606460: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer_plugin.so.7'; dlerror: libnvinfer_plugin.so.7: cannot open shared object file: No such file or directory
2023-05-31 10:50:35.606470: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Cannot dlopen some TensorRT libraries. If you would like to use Nvidia GPU with TensorRT, please make sure the missing libraries mentioned above are installed properly.
>>> tf.config.list_physical_devices('GPU')
2023-05-31 10:50:56.688374: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudnn.so.8'; dlerror: libcudnn.so.8: cannot open shared object file: No such file or directory
2023-05-31 10:50:56.688462: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1934] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.
Skipping registering GPU devices...

解决方案（跟着走）

• 查看cuda, cudnn和tensorflow对应关系：https://www.tensorflow.org/install/source#tested_build_configurations
• 通过'nvidia-smi'查看当前服务器cuda版本（当前是11.7），所以在选择tensorflow的版本时要看cuda最低要求，不能超过服务器cuda version
• 官网安装步骤（换成英文页面）：https://www.tensorflow.org/install/pip 。在step 4 GPU setup需要按照自己的cuda版本修改cudatoolkit（我是=cuda版本）和nvidia-cudnn-cu11的版本
• https://github.com/tensorflow/tensorflow/issues/57679 （找到id为'didamain'）

问题2 （还未解决）

• 描述：按照之前的步骤成功安装tf之后，启动jupyter lab，一用GPU训练就kernel 重启。
• 
• 按照官方代码：https://www.tensorflow.org/tutorials/keras/classification 。 出现相同问题说明不是代码的问题，确实是tensorflow有问题。

解决 and 终极安装

• note: 要正确选择 cudatoolkit、 cudnn（conda安装就是cudnn，pip安装叫nvidia-cudnn-cu11）和 tensorflow的版本组合，有不少组合并不可用，尽管满足tensorflow给出的对应关系。
• 经实验可用的组合有：
  • py3.8+tf2.9+cuda11.3.1+cudnn8.2.1
  • py3.8+tf2.9+cuda11.6+cudnn8.6
• 安装步骤依旧按照tensorflow官网：https://www.tensorflow.org/install/pip 。其中安装cudatoolkit, cudnn, tensorflow需要改（用conda 或者 pip，conda官方的源还是清华的源，还得看清华的源有没有该版本资源）
• 我安装上述三个组件的指令：
  • conda install cudatoolkit=11.3.1
  • conda install cudnn=8.2.1 (没有在pip找到该版本)
  • pip install tensorflow==2.9.*
• ubuntu22.04 需要额外一步
  

最新参考目录：https://blog.csdn.net/LIWEI940638093/article/details/113811563