(多圖)CentOS下安裝NVIDIA driver的超詳細教程
NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.錯誤及解決方式
前言
這個錯誤是筆者在CentOS 7下使用nvidia-smi
指令時所出現的錯誤。
錯誤訊息
NVIDIA-SMI has failed because it couldn’t communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.
錯誤訊息說明瞭NVIDIA driver並沒有安裝。
解決方法
解決方法就是安裝它所需要的nvidia driver,具體步驟可以參照這個連結:How to install Nvidia-Docker 2.0 on CentOS 7.md
具體步驟
更新作業系統
$ yum -y update
$ yum -y install kernel-devel kernel-headers gcc make
$ yum -y install dkms epel-release
$ shutdown -r now
Disable old Nouveau driver
$ cat > /etc/modprobe.d/blacklist.conf << EOF
blacklist nouveau
options nouveau modeset=0
EOF
$ mv /boot/initramfs-`uname -r`.img /boot/initramfs-`uname -r`.img.bak
$ wget http://mirror.centos.org/centos/6/os/x86_64/Packages/busybox-1.15.1-21.el6_6.x86_64.rpm
$ rpm -Uvh busybox-1.15.1-21.el6_6.x86_64.rpm
$ yum install -y ntfs-3g
$ dracut -v /boot/initramfs-`uname -r`.img `uname -r`
安裝Nvidia driver
獲取Nvidia driver:
$ yum install -y wget
$ wget http://us.download.nvidia.com/XFree86/Linux-x86_64/390.67/NVIDIA-Linux-x86_64-390.67.run \
-O NVIDIA-Linux-x86_64-390.67.run
註:欲獲取最新版的nvidia driver可將上述連結改為:http://us.download.nvidia.com/XFree86/Linux-x86_64/410.78/NVIDIA-Linux-x86_64-410.78.run。
使它變為可執行檔,然後執行:
$ chmod +x NVIDIA-Linux-x86_64-390.67.run
$ ./NVIDIA-Linux-x86_64-390.67.run
此時會出現如下輸出:
接著會進入以下畫面:
筆者這裡是重新安裝nvidia driver,所以畫面可能不太一樣。
是否使用DKMS建構及安裝kernel module?這裡選"是"。
只有一個OK,當然是按Enter。
是否安裝與32-bit相容的library?這裡選"是"。
重新安裝libglvnd。
建構及安裝DKMS kernel module,這得花上幾分鐘的時間。
是否執行nvidia-xconfig utility?這裡選"是"。
X configuration file成功更新。Nvidia driver安裝成功。
測試Nvidia driver
使用以下指令檢查安裝是否成功:
$ nvidia-smi
參考連結
How to install Nvidia-Docker 2.0 on CentOS 7.md
wangruohui/Install NVIDIA Driver and CUDA.md
Chapter 4. Installing the NVIDIA Driver
Chapter 6. Configuring X for the NVIDIA Driver
Setup NVIDIA drivers on Ubuntu and derivates
INSTALLING NVIDIA DRIVERS ON RHEL OR CENTOS 7