如何在ubuntu20.04中安装cuDNNv8.0.5

要在Ubuntu 20.04系统上安装CUDA 11.1 (通常包含了cuDNN v8.0.5)并配置TensorFlow支持，你可以按照以下步骤操作：

1. **更新包列表**:

   sudo apt update && sudo apt upgrade

2. **安装依赖库**:

   sudo apt install -y build-essential checkinstall libncurses5-dev libncursesw5-dev libcurl4-openssl-dev libssl-dev libreadline6-dev libdb5.3-dev libgdbm-dev libsqlite3-dev tk-dev libx11-dev libelf-dev libbz2-dev liblzma-dev liblz4-tool liblz4-dev libzstd-dev libopenblas-dev liblapack-dev libatlas-base-dev gfortran

3. **下载CUDA**:
首先访问CUDA官网 (https://developer.nvidia.com/cuda-downloads)，选择适合Ubuntu 20.04的版本(例如CUDA 11.1)，然后下载DEB包。

4. **安装CUDA**:
下载完成后，解压并安装CUDA，例如对于`.deb`文件：

   sudo dpkg -i cuda-repo-ubuntu2004_11.1.0-455.21.04_amd64.deb
   sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/repos/ubuntu/repos/lists/xenial/xenial.txt
   sudo apt update
   sudo apt-get install cuda

5. **安装cuDNN**:
CUDA安装会自动处理cuDNN的安装，如果你需要手动安装，首先从NVIDIA Developer网站(https://developer.nvidia.com/cudnn)下载cuDNN压缩包，解压后运行类似下面的命令：

   sudo cp -r /path/to/cudnn/include /usr/local/cuda/include/

6. **配置环境变量**:
将CUDA路径添加到`LD_LIBRARY_PATH`和`PYTHONPATH`:

   echo "export PATH=/usr/local/cuda/bin:$PATH" >> ~/.bashrc
   export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH
   export PYTHONPATH=/usr/local/cuda/include/python3.7:/usr/local/cuda:$PYTHONPATH
   source ~/.bashrc

7. **验证安装**:
运行一些CUDA示例程序或检查CUDA、cuDNN是否可用：

   nvidia-smi
   python3 -c "import tensorflow as tf; print(tf.test.is_gpu_available())"