Centos7 KVM 虚拟机加载NVIDIA驱动
文章目录1 Centos7.7宿主机装NVIDIA驱动2 KVM虚拟机安装NVIDIA driver1 Centos7.7宿主机装NVIDIA驱动NVIDIA驱动官方下载地址# 1 关闭nouveau driverecho "blacklist nouveau" >> /etc/modprobe.d/blacklist.confcat >>/lib/modprobe.d/d
·
1 Centos7.7宿主机安装NVIDIA驱动
（1）关闭nouveau driver
echo "blacklist nouveau" >> /etc/modprobe.d/blacklist.conf
cat >>/lib/modprobe.d/dist-blacklist.conf<<EOF
blacklist nouveau
options nouveau modeset=0
EOF
（2）给当前镜像做备份，并建立新的镜像
mv /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r)-nouveau.img
dracut /boot/initramfs-$(uname -r).img $(uname -r)
reboot # 重启机器
（3）安装NVIDIA驱动
rpm -ivh NVIDIA-vGPU-rhel-7.7-460.32.04.x86_64.rpm
reboot # 安装完后重启系统
（4）验证系统是否加载NVIDIA vGPU驱动
[root@localhost NVIDIA]# lsmod | grep vfio
nvidia_vgpu_vfio 50250 19
nvidia 34001825 295 nvidia_vgpu_vfio
vfio_mdev 12841 1
mdev 20336 2 vfio_mdev,nvidia_vgpu_vfio
vfio_iommu_type1 22440 1
vfio 32657 6 vfio_mdev,nvidia_vgpu_vfio,vfio_iommu_type1
（5）nvidia-smi验证
[root@localhost NVIDIA]# nvidia-smi
Tue May 11 15:51:09 2021
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 460.32.04 Driver Version: 460.32.04 CUDA Version: N/A |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 Tesla P40 On | 00000000:3D:00.0 Off | 0 |
| N/A 19C P8 19W / 250W | 41MiB / 23039MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
| 1 Tesla P40 On | 00000000:41:00.0 Off | 0 |
| N/A 20C P8 19W / 250W | 22899MiB / 23039MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 1 N/A N/A 5268 C+G vgpu 22848MiB |
+-----------------------------------------------------------------------------+
2 KVM虚拟机安装NVIDIA 驱动
（1）配置kvm虚拟机网络跟宿主机桥接
# 关闭NetworkManager服务
systemctl stop NetworkManager
systemctl disable NetworkManager
# 创建新的br0网卡，跟物理网卡进行绑定
virsh iface-bridge enp97s0f0 br0
（2）网卡验证
[root@localhost ~]# brctl show
bridge name bridge id STP enabled interfaces
br0 8000.b4055d3f1e78 yes enp97s0f0
vnet0
virbr0 8000.525400789520 yes virbr0-nic
（3）拆分物理GPU，创建vGPU
# 通过nvidia-smi查看显卡bus_id，以 0000:3d:00.0 为例
cd /sys/class/mdev_bus/0000:3d:00.0/mdev_supported_types
（4）列出物理GPU可以拆分的类型
[root@localhost ~]# for i in nvidia-* ; do echo -n "$i "; cat $i/name ; done
nvidia-156 GRID P40-2B
nvidia-215 GRID P40-2B4
nvidia-241 GRID P40-1B4
nvidia-283 GRID P40-4C
nvidia-284 GRID P40-6C
nvidia-285 GRID P40-8C
nvidia-286 GRID P40-12C
nvidia-287 GRID P40-24C
nvidia-46 GRID P40-1Q
nvidia-47 GRID P40-2Q
nvidia-48 GRID P40-3Q
nvidia-49 GRID P40-4Q
nvidia-50 GRID P40-6Q
nvidia-51 GRID P40-8Q
nvidia-52 GRID P40-12Q
nvidia-53 GRID P40-24Q
nvidia-54 GRID P40-1A
nvidia-55 GRID P40-2A
nvidia-56 GRID P40-3A
nvidia-57 GRID P40-4A
nvidia-58 GRID P40-6A
nvidia-59 GRID P40-8A
nvidia-60 GRID P40-12A
nvidia-61 GRID P40-24A
nvidia-62 GRID P40-1B
（5）这里我用的是 Q 渲染模式
# 详细的模式选择可参考官方文档
# https://docs.nvidia.com/grid/10.0/grid-vgpu-user-guide/index.html#virtual-gpu-types-grid
[root@localhost mdev_supported_types]# cd nvidia-53
[root@localhost nvidia-53]# ls
available_instances create description device_api devices namels
[root@localhost nvidia-53]# cat available_instances
1
# 为vGPU生成唯一标识符
[root@localhost nvidia-53]# uuidgen
3d107946-44ef-4c87-aff2-dff273402208
[root@localhost nvidia-53]# echo "3d107946-44ef-4c87-aff2-dff273402208" > create
# 确认vGPU被创建
[root@localhost ~]# ls -l /sys/bus/mdev/devices/
total 0
lrwxrwxrwx. 1 root root 0 Apr 30 15:44 3d107946-44ef-4c87-aff2-dff273402208 -> ../../../devices/pci0000:3a/0000:3a:00.0/0000:3b:00.0/0000:3c:14.0/0000:41:00.0/3d107946-44ef-4c87-aff2-dff273402208
（6）kvm虚拟机引用此vGPU
[root@localhost ~]# virsh list --all
Id Name State
----------------------------------------------------
- centos7.0 shut off
[root@localhost ~]# virsh edit centos7.0
# 添加如下字段
<hostdev mode='subsystem' type='mdev' managed='no' model='vfio-pci' display='off'>
<source>
<address uuid='3d107946-44ef-4c87-aff2-dff273402208'/>
</source>
<address type='pci' domain='0x0000' bus='0x00' slot='0x0a' function='0x0'/>
</hostdev>
# 启动kvm虚拟机
[root@localhost ~]# virsh start centos7.0
（7）拷贝NVIDIA的.run程序，这个可执行文件在官网下载的zip包中被一并打包
[root@aster NVIDIA]# ls
NVIDIA-Linux-x86_64-460.32.03-grid.run
# 关闭nouveau driver
echo "blacklist nouveau" >> /etc/modprobe.d/blacklist.conf
cat >>/lib/modprobe.d/dist-blacklist.conf<<EOF
blacklist nouveau
options nouveau modeset=0
EOF
# 给当前镜像做备份，并建立新的镜像
mv /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r)-nouveau.img
dracut /boot/initramfs-$(uname -r).img $(uname -r)
reboot # 重启机器
（8）安装NVIDIA driver，需要在init3模式下
[root@aster NVIDIA]# init 3
[root@aster NVIDIA]# chmod 777 NVIDIA-Linux-x86_64-440.107-grid.run
[root@aster NVIDIA]# bash NVIDIA-Linux-x86_64-440.107-grid.run
[root@aster NVIDIA]# reboot # 安装完之后reboot
# 重启后查看
[root@aster NVIDIA]# nvidia-smi
Tue May 11 16:36:51 2021
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 460.32.03 Driver Version: 460.32.03 CUDA Version: 11.2 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 GRID P40-24Q On | 00000000:00:0A.0 Off | N/A |
| N/A N/A P8 N/A / N/A | 3636MiB / 24576MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+
| Series | Optimal Workload |
|---|---|
| Q-series | Virtual workstations for creative and technical professionals who require the performance and features of Quadro technology |
| C-series | Compute-intensive server workloads, such as artificial intelligence (AI), deep learning, or high-performance computing (HPC) |
| B-series | Virtual desktops for business professionals and knowledge workers |
| A-series | App streaming or session-based solutions for virtual applications users |
华为开发者空间，是为全球开发者打造的专属开发空间，汇聚了华为优质开发资源及工具，致力于让每一位开发者拥有一台云主机，基于华为根生态开发、创新。
更多推荐
3
0- 0
已为社区贡献6条内容
所有评论(0)