linux Ubuntu 20.04.6 + pytroch1.13.1 + cudnn8.4.0 + cudatoolkit11.6.0 python3.7 + Anaconda3-5.3.1

驱动

安装cudatoolkit11.6.0

下载地址 https://developer.nvidia.com/cuda-toolkit-archive
sh  cuda_11.6.0_510.39.01_linux.run

安装cudnn8.4.0

下载地址 https://developer.nvidia.com/rdp/cudnn-archive
tar -xvf cudnn-linux-x86_64-8.4.0.27_cuda11.6-archive .tar.xz #解压
sudo cp cudnn-linux-x86_64-8.4.0.27_cuda11.6-archive/include/cudnn*.h /usr/local/cuda11.6/include
sudo cp  cudnn-linux-x86_64-8.4.0.27_cuda11.6-archive/lib64/libcudnn* /usr/local/cuda-11.6/lib64
sudo chmod a+r /usr/local/cuda/include/cudnn*.h /usr/local/cuda11.6/lib64/libcudnn*

安装anaconda

wget https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/Anaconda3-5.3.1-Linux-x86_64.sh
sh Anaconda3-5.3.1-Linux-x86_64.sh

创建虚拟环境

conda create -n pytorch python=3.7

此代码含义是python版本3.7 虚拟环境名字是pytorch

下载pytorch1.13.1

pip install torch==1.13.1+cu116 torchvision==0.14.1+cu116 torchaudio==0.13.1 --extra-index-url https://download.pytorch.org/whl/cu116

配置环境

export PATH=$PATH:/usr/local/cuda-11.6/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-11.6/lib64
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-11.6/extras/CUPTI/lib64
export PATH="/usr/local/cuda-11.6/bin:$PATH"
export LD_LIBRARY_PATH="/usr/local/cuda-11.6/lib64:$LD_LIBRARY_PATH"
export PATH="/home/*/.local/bin:$PATH/"
export PATH="/usr/bin:$PATH"

sourse .bashrc

附上最难下载的cudnn资源密码123456