Ubuntu安装机器学习相关环境

说明：针对Ubuntu 18.04 LTS，Nvidia 3080Ti环境

注意随着驱动版本更新，510驱动可能无法适配最新的CUDA，适配表参考https://docs.nvidia.com/cuda/cuda-toolkit-release-notes/index.html

sudo apt install nvidia-driver-510 --fix-missing

· https://developer.nvidia.com/cuda-toolkit-archive 选择CUDA11.3后，选择Linux – x86_64 – Ubuntu - 20.04 – deb(network)可以看到如下安装代码，按顺序执行即可
· 注意上面需要根据pytorch对CUDA版本进行选择：查看https://pytorch.org/get-started/locally/找支持的最新CUDA版本，当前为CUDA 11.3

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/cuda-ubuntu1804.pin
sudo mv cuda-ubuntu1804.pin /etc/apt/preferences.d/cuda-repository-pin-600
sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/3bf863cc.pub
sudo add-apt-repository "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/ /"
sudo apt-get update
sudo apt-get -y install cuda-11-3 # 若不设置-11-3则默认安装当前版本CUDA

· 下面验证安装成功

nvcc -V
nvidia-smi
python
# >>> import torch
# >>> print(torch.cuda.is_available())

在https://pytorch.org/get-started/locally/选择stable-linux-pip-python-CUDA11.3即可

python和python3的ln连接修改为ubuntu自带的python3.6，参考连接删除错误包
https://phoenixnap.com/kb/fix-sub-process-usr-bin-dpkg-returned-error-code-1

https://stackoverflow.com/questions/56431461/how-to-remove-cuda-completely-from-ubuntu