服务器非root配置anaconda+cuda11.8+cudnn

安装Anaconda

• 【安装Anaconda后出现的conda: command not found问题】

安装cuda11.8

wget https://developer.download.nvidia.com/compute/cuda/11.8.0/local_installers/cuda_11.8.0_520.61.05_linux.run
sudo sh cuda_11.8.0_520.61.05_linux.run

如果没有管理员权限

bash
sh cuda_11.8.0_520.61.05_linux.run

安装完成进入向导 只选择cudatoolkit11.8，然后更改cuda的安装目录建议是/home/yourname/CUDA/CUDA11.8 同时更改lib安装路径/home/yourname/CUDA/CUDA11.8/mylib

修改变量

vim /.bashrc

在CUDA initial下面添加

#export PATH=/usr/local/cuda–10.1/bin:PATH
#export LD_LIBRARYPATH=$LD_LIBRARY_PATH:/usr/local/cuda–10.1/lib64：$LD_LIBRARY_PATH
export CUDA_HOME="/home/SJU/jyh/CUDA"
#export PATH="$CUDA_HOME/CUDA11.2/bin:$PATH"
#export LD LIBRARY_PATH="$CUDA_HOME/CUDA11.2/lib64:$CUDA_HOME/CUDA11.2/mylib/lib64:$LD_LIBRARY_PATH"
export PATH="$CUDA_HOME/CUDA11.8/bin:$PATH"
export LD_LIBRARY_PATH="$CUDA_HOME/CUDA11.8/lib64：$CUDA_HOME/CUDA11.8/mylib/1ib64:$LD_LIBRARY_PATH"

启用配置

source ~/.bashrc

验证

nvcc –V

安装cudnn

前往英伟达官网，需要账号 下载后上传

tar –xf cudnn–linux–x86_64–8.9.7.29_cuda11–archive.tar.xz

cd cudnn–linux–x86_64–8.9.7.29_cuda11–archive
cp ./include/cudnn*.h /home/STU/jyh/CUDA/CUDA11.8/include
cp ./lib/libcudnn* /home/STU/jyh/CUDA/CUDA11.8/lib64

chmod a+r /home/STU/jyh/CUDA/CUDA11.8/include/cudnn*.h /home/STU/jyh/CUDA/CUDA11.8/lib64/libcudnn*

参考

• 【Linux服务器上安装CUDA11.2和对应的cuDNN 8.4.0】
• 【Linux非root权限用户安装CUDA和CUDNN】

附录

常用命令

nvidia-smi
nvcc -V
vim ~/.bashrc
source ~/.bashrc