Manually Installing a Tesla Driver on a GPU-accelerated ECS

Scenarios

Before using a GPU-accelerated ECS, make sure that the desired Tesla driver and CUDA toolkit have been installed on the ECS for computing acceleration.

A computing-accelerated (P series) ECS created using a public image has had a Tesla driver of a specified version installed by default.
After a GPU-accelerated ECS is created using a private image, it must have a Tesla driver installed. Otherwise, computing acceleration will not take effect.

This section describes how to install a Tesla driver and CUDA toolkit on a GPU-accelerated ECS.

Notes

The ECS must have an EIP bound.
Check whether the CUDA toolkit and Tesla driver have been installed on the ECS.

If the CUDA toolkit has not been installed, download it from the official NVIDIA website and install it. A Tesla driver matching the CUDA version will be automatically installed then. However, if there are specific requirements or dependencies on the Tesla driver version, download the matching Tesla driver from the official NVIDIA website first and then install the driver before installing the CUDA toolkit.
If a Tesla driver has been installed on the ECS, check the driver version. Before installing a new driver version, uninstall the original Tesla driver to prevent an installation failure due to driver conflicts.

Installation process:

Obtaining a Tesla Driver and CUDA Toolkit
Installing a Tesla Driver
- Installing a Tesla Driver on a Linux ECS
Installing a CUDA Toolkit
- Installing the CUDA Toolkit on a Linux ECS

Installing a Tesla Driver on a Linux ECS

The following uses Ubuntu 16.04 64bit as an example to describe how to install the Tesla driver matching CUDA 10.1 on a GPU-accelerated ECS.

The Linux kernel version is compatible with the driver version. If installing the driver failed, check the driver installation log, which is generally stored in /var/log/nvidia-installer.log. If the log shows that the failure was caused by a driver compilation error, for example, the get_user_pages parameter setting is incorrect, the kernel version is incompatible with the driver version. In such a case, select the desired kernel version and driver version and reinstall them. It is recommended that the release time of the kernel version and driver version be the same.

Log in to the ECS.
Update the system software based on the OS.
- Ubuntu
  Update the software installation source: apt-get -y update
  
  Install necessary programs: apt-get install gcc g++ make
- CentOS
  Update the software installation source: yum -y update --exclude=kernel* --exclude=centos-release* --exclude=initscripts*
  
  Install the desired program: yum install -y kernel-devel-`uname -r` gcc gcc-c++
Download the NVIDIA driver package.
Select a driver version at NVIDIA Driver Downloads based on the ECS type. Click SEARCH.

Figure 1 Selecting a NVIDIA driver version
Select a driver version as required. The following uses Tesla 418.67 as an example.
Figure 2 Selecting a driver version
Click the driver to be downloaded. On the TESLA DRIVER FOR LINUX X64 page that is displayed, click DOWNLOAD.
Copy the download link.
Figure 3 Copying the download link
Run the following command on the ECS to download the driver:
wget Copied link

For example, wget http://us.download.nvidia.com/tesla/418.67/NVIDIA-Linux-x86_64-418.67.run

Figure 4 Obtaining the installation package
Run the following command to install the driver:
sh NVIDIA-Linux-x86_64-418.67.run
(Optional) If the following information is displayed after the command for installing the driver is executed, disable the Nouveau driver.
Figure 5 Disabling the Nouveau driver
1. Run the following command to check whether the Nouveau driver has been installed:
  lsmod | grep nouveau
  - If the command output contains information about the Nouveau driver, the Nouveau driver has been installed and must be disabled. Then, go to step 9.b.
  - If the command output does not contain information about the Nouveau driver, the Nouveau driver has been disabled. Then, go to step 10.
2. Edit the blacklist.conf file.
  If the /etc/modprobe.d/blacklist.conf file is unavailable, create it.
  
  vi /etc/modprobe.d/blacklist.conf
  
  Add the following statement to the end of the file:
```
blacklist nouveau
options nouveau modeset=0
```
1. Run the following command to back up and create an initramfs application:
  - Ubuntu
    sudo update-initramfs -u
  - CentOS:
    mv /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r).img.bak
    
    dracut -v /boot/initramfs-$(uname -r).img $(uname -r)
1. Restart the ECS:
  reboot
Select OK for three consecutive times as prompted to complete the driver installation.
Figure 6 Completing the NVIDIA driver installation
Run the following command to set systemd:
systemctl set-default multi-user.target
Run the reboot command to restart the ECS.
Log in to the ECS and run the nvidia-smi command. If the command output contains the installed driver version, the driver has been installed.
Figure 7 Viewing the NVIDIA driver version

Installing the CUDA Toolkit on a Linux ECS

The following uses Ubuntu 16.04 64bit as an example to describe how to install the CUDA 10.1 toolkit on a GPU-accelerated ECS.

Log in to the ECS.
Update the system software based on the OS.
- Ubuntu
  Update the software installation source: apt-get -y update
  
  Install necessary programs: apt-get install gcc g++ make
- CentOS
  Update the software installation source: yum -y update --exclude=kernel* --exclude=centos-release* --exclude=initscripts*
  
  Install the desired program: yum install -y kernel-devel-`uname -r` gcc gcc-c++
On the CUDA download page, set parameters according to the information shown in Obtaining a Tesla Driver and CUDA Toolkit.
Figure 8 Selecting a CUDA version
Find the link for downloading CUDA 10.1 and copy the link.
Figure 9 Copying the link for downloading CUDA

Run the following command on the ECS to download CUDA:
wget Copied link

For example, wget https://developer.nvidia.com/compute/cuda/10.1/Prod/local_installers/cuda_10.1.105_418.39_linux.run

Figure 10 Downloading CUDA

Install CUDA.
Follow the instructions provided on the official NVIDIA website.
Figure 11 Installing CUDA
Run the following command to install CUDA:
sh cuda_10.1.243_418.87.00_linux.run
Select accept on the installation page and press Enter.
Figure 12 Installing CUDA_1
Select Install and press Enter to start the installation.
Figure 13 Installing CUDA_2

Figure 14 Completing the installation
Run the following command to switch to /usr/local/cuda-10.1/samples/1_Utilities/deviceQuery:
cd /usr/local/cuda-10.1/samples/1_Utilities/deviceQuery
Run the make command to automatically compile the deviceQuery program.
Run the following command to check whether CUDA has been installed:
./deviceQuery

If the command output contains the CUDA version, CUDA has been installed.

Figure 15 deviceQuery common output
Check the CUDA version.
/usr/local/cuda/bin/nvcc -V

Figure 16 Checking the CUDA version
Run the following command to enable the persistent mode:
sudo nvidia-smi -pm 1

Enabling the persistent mode optimizes the GPU performance on Linux ECSs.