Using CentOS 7.7 and a Tesla P100 GPU as an example.
1. Preparing the Base Environment
- Install the lspci command
1
| yum install -y pciutils
|
- Check whether the GPU supports CUDA
1
2
3
| lspci | grep -i nvidia
00:09.0 3D controller: NVIDIA Corporation GP100GL [Tesla P100 PCIe 12GB] (rev a1)
|
List of CUDA-capable GPUs: https://developer.nvidia.com/cuda-gpus
- Check whether the system supports CUDA
1
2
3
4
| uname -m && cat /etc/redhat-release
x86_64
CentOS Linux release 7.7.1908 (Core)
|
List of CUDA-capable OSes: https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements
- Install system tooling packages
1
2
3
| yum update -y
yum install -y wget vim gcc
yum install kernel-devel-$(uname -r) kernel-headers-$(uname -r)
|
Version 19.03 or later is required, reference link .
Reference for installing Docker: Installing a Specific Docker Version .
2. Installing the GPU Driver & CUDA
2.1 Disabling the System’s Default nouveau Driver
Before blacklisting:
1
2
3
4
5
6
7
8
9
10
| lsmod | grep nouveau
nouveau 1898794 0
mxm_wmi 13021 1 nouveau
wmi 21636 2 mxm_wmi,nouveau
video 24538 1 nouveau
i2c_algo_bit 13413 1 nouveau
ttm 96673 2 bochs_drm,nouveau
drm_kms_helper 186531 2 bochs_drm,nouveau
drm 456166 5 ttm,bochs_drm,drm_kms_helper,nouveau
|
Disabling nouveau :
1
2
| bash -c "echo blacklist nouveau > /etc/modprobe.d/blacklist-nvidia-nouveau.conf"
bash -c "echo options nouveau modeset=0 >> /etc/modprobe.d/blacklist-nvidia-nouveau.conf"
|
Rebuild the initramfs image
1
2
| mv /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r).img.bak
dracut /boot/initramfs-$(uname -r).img $(uname -r)
|
Reboot the system; after blacklisting:
1
2
3
| lsmod | grep nouveau
(no output)
|
2.2 Installing the GPU Driver
There are two installation methods:
- The first: install the kmod-nvidia driver
Add the repository
1
2
| rpm --import https://www.elrepo.org/RPM-GPG-KEY-elrepo.org
rpm -Uvh http://www.elrepo.org/elrepo-release-7.0-2.el7.elrepo.noarch.rpm
|
Install nvidia-detect :
1
| yum install -y nvidia-detect
|
Check whether a matching kmod-nvidia version is available:
Install the kmod-nvidia driver:
1
| yum install -y kmod-nvidia
|
- The second: download the driver from the official site and install it
On the Nvidia official driver download page, find the GPU type shown by the lspci | grep -i nvidia command.
1
2
| wget http://cn.download.nvidia.com/tesla/440.64.00/nvidia-driver-local-repo-rhel7-440.64.00-1.0-1.x86_64.rpm
rpm -Uvh nvidia-driver-local-repo-rhel7-440.64.00-1.0-1.x86_64.rpm
|
You can also download the Shell script to install
1
2
3
| wget http://us.download.nvidia.com/tesla/440.33.01/NVIDIA-Linux-x86_64-440.64.00.run
chmod +x NVIDIA-Linux-x86_64-440.64.00.run
bash ./NVIDIA-Linux-x86_64-440.64.00.run
|
2.3 Installing CUDA
On the Nvidia developer cuda-toolkit-archive page, find the latest toolkit version. Following the page’s prompts, select your operating system; below are the installation commands obtained for CentOS 7.7:
1
2
3
4
5
| wget http://developer.download.nvidia.com/compute/cuda/10.2/Prod/local_installers/cuda-repo-rhel7-10-2-local-10.2.89-440.33.01-1.0-1.x86_64.rpm
sudo rpm -i cuda-repo-rhel7-10-2-local-10.2.89-440.33.01-1.0-1.x86_64.rpm
sudo yum clean all
sudo yum -y install nvidia-driver-latest-dkms cuda
sudo yum -y install cuda-drivers
|
2.4 Verifying the Installation
After rebooting the machine, check whether Nvidia CUDA was installed successfully.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
| nvidia-smi
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 440.82 Driver Version: 440.82 CUDA Version: 10.2 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 Tesla P100-PCIE... Off | 00000000:00:09.0 Off | 0 |
| N/A 35C P0 27W / 250W | 0MiB / 12198MiB | 6% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+
|
3. Installing nvidia-docker
nvidia-docker provides support for using GPU acceleration inside Docker.
1
2
3
| distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.repo | sudo tee /etc/yum.repos.d/nvidia-docker.repo
yum install -y nvidia-container-runtime nvidia-container-toolkit nvidia-docker2
|
Edit the /etc/docker/daemon.json file and add the following content:
1
2
3
4
5
6
7
8
| {
"runtimes": {
"nvidia": {
"path": "/usr/bin/nvidia-container-runtime",
"runtimeArgs": []
}
}
}
|
- Restart Docker for the change to take effect
1
| systemctl restart docker
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
| docker run --gpus all nvidia/cuda:10.0-base nvidia-smi
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 440.82 Driver Version: 440.82 CUDA Version: 10.2 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 Tesla P100-PCIE... Off | 00000000:00:09.0 Off | 0 |
| N/A 36C P0 26W / 250W | 0MiB / 12198MiB | 6% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
| docker run --runtime=nvidia nvidia/cuda:10.0-base nvidia-smi
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 440.82 Driver Version: 440.82 CUDA Version: 10.2 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 Tesla P100-PCIE... Off | 00000000:00:09.0 Off | 0 |
| N/A 36C P0 26W / 250W | 0MiB / 12198MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
| nvidia-docker run nvidia/cuda:10.0-base nvidia-smi
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 440.82 Driver Version: 440.82 CUDA Version: 10.2 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 Tesla P100-PCIE... Off | 00000000:00:09.0 Off | 0 |
| N/A 35C P0 26W / 250W | 0MiB / 12198MiB | 6% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+
|
4. References
