NVIDIA驱动升级与CUDA多版本管理

版本对应参考信息:

Step 1: 更新NVIDIA驱动

sudo init 3            # 进入字符模式
sudo nvidia-uninstall  # 卸载已有驱动
sudo ./NVIDIA-Linux-x86_64-410.129-diagnostic.run -no-x-check -no-nouveau-check -no-opengl-files # 安装新驱动

可能报错信息与对应解决办法

[1] sudo: ./NVIDIA-Linux-x86_64-418.43.run: command not found

sudo chmod +x NVIDIA-Linux-x86_64-418.43.run   # 未赋予驱动写的权限
sudo ./NVIDIA-Linux-x86_64-418.43.run

[2] An NVIDIA kernel module ‘nvidia’ appears to already be loaded in your kernel

sudo lsof -n -w /dev/nvidia*  # 查到占用进程号PID, sudo kill -9 PID

[3] An NVIDIA kernel module ‘nvidia-uvm’ appears to already be loaded in your kernel.

sudo lsof | grep nvidia.uvm   # 查到占用进程号PID, sudo kill -9 PID

Step 2: 安装新版CUDA

# 安装依赖库，否则会出现'Missing recommended library'错误
sudo apt-get install freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev

# 赋予文件读写权限
sudo chmod +x cuda_10.0.130_410.48_linux.run
# 开始安装CUDA
sudo ./cuda_10.0.130_410.48_linux.run

#### 安装过程