P40 GPU 直透
####pre
1.vsphere6.5 平台 + 4 张P40 GPU卡
2.guest os nvidia 驱动
###1.uiweb上设置 gpu 直通 模式
设置好后,需要重新引导系统,重启后,如上图,直通状态为 活动
###2. 命令行进主机系统,查看
[root@gpu52:/etc/vmware] cat /etc/vmware/esx.conf | grep passthru
/device/00000:136:00.0/owner = "passthru"
/device/00000:005:00.0/owner = "passthru"
/device/00000:008:00.0/owner = "passthru"
/device/00000:131:00.0/owner = "passthru"
说明,4张gpu卡已设置直透了
###3. 新增2台虚拟机win2K12 和 test-3 (centos7.8)
###4.使用vcenter操作
#4.1
#4.2 选择虚拟机–右键–编辑设置–虚拟机硬件–添加PCI设备
#4.3 选择PCI设备—预留对等内存–
#4.4 虚拟机选项—高级 —编辑配置
新增
pciPassthru.use64bitMMIO TRUE
pciPassthru.64bitMMIOSizeGB 64
##4.5
####5.虚机安装显卡驱动
#5.1 下载
####5.2 centos7 安装驱动
####查看显卡信息
[root@localhost ~]# lspci | grep -i NVIDIA
0b:00.0 3D controller: NVIDIA Corporation GP102GL [Tesla P40] (rev a1)
####安装依赖环境
yum install kernel-devel gcc -y
###.检查内核版本和源码版本
[root@localhost ~]# ls /boot | grep vmlinu
vmlinuz-0-rescue-07cf47a5080d46b8a3558c3c1979b0ec
vmlinuz-3.10.0-1127.el7.x86_64
[root@localhost ~]#
[root@localhost ~]# rpm -aq | grep kernel-devel
kernel-devel-3.10.0-1127.19.1.el7.x86_64
###屏蔽系统自带的nouveau
[root@localhost ~]# lsmod | grep nouveau
nouveau 1898794 0
video 24538 1 nouveau
mxm_wmi 13021 1 nouveau
wmi 21636 2 mxm_wmi,nouveau
i2c_algo_bit 13413 1 nouveau
drm_kms_helper 186531 2 nouveau,vmwgfx
ttm 96673 2 nouveau,vmwgfx
drm 456166 5 ttm,drm_kms_helper,nouveau,vmwgfx
修改 /lib/modprobe.d/dist-blacklist.conf
将nvidiafb注释掉:
#blacklist nvidiafb
然后添加以下语句:
blacklist nouveau
options nouveau modeset=0
####重建initramfs image
[root@localhost ~]# mv /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r).img.bak
[root@localhost ~]#
[root@localhost ~]# dracut /boot/initramfs-$(uname -r).img $(uname -r)
####修改运行级别为文本模式
[root@localhost ~]# systemctl set-default multi-user.target
Removed symlink /etc/systemd/system/default.target.
Created symlink from /etc/systemd/system/default.target to /usr/lib/systemd/system/multi-user.target.
####,重启,
[root@localhost ~]# lsmod | grep nouveau
[root@localhost ~]#
[root@localhost ~]#
[root@localhost ~]#
[root@localhost ~]#
####上传 驱动
[root@localhost ~]# ls
anaconda-ks.cfg NVIDIA-GRID-vSphere-6.5-430.99-432.44.zip
[root@localhost ~]#
[root@localhost ~]# unzip -d /tmp NVIDIA-GRID-vSphere-6.5-430.99-432.44.zip
Archive: NVIDIA-GRID-vSphere-6.5-430.99-432.44.zip
inflating: /tmp/430.99-432.44-grid-gpumodeswitch-user-guide.pdf
inflating: /tmp/430.99-432.44-grid-licensing-user-guide.pdf
inflating: /tmp/430.99-432.44-grid-software-quick-start-guide.pdf
inflating: /tmp/430.99-432.44-grid-vgpu-release-notes-vmware-vsphere.pdf
inflating: /tmp/430.99-432.44-grid-vgpu-user-guide.pdf
inflating: /tmp/430.99-432.44-whats-new-vgpu.pdf
inflating: /tmp/432.44_grid_win10_server2016_server2019_64bit_international.exe
inflating: /tmp/432.44_grid_win7_win8_server2008R2_server2012R2_64bit_international.exe
inflating: /tmp/NVD.NVIDIA_bootbank_NVIDIA-VMware_430.99-1OEM.650.0.0.4598673-offline_bundle-16285628.zip
inflating: /tmp/NVIDIA-Linux-x86_64-430.99-grid.run
inflating: /tmp/NVIDIA-VMware-430.99-1OEM.650.0.0.4598673.x86_64.vib
[root@localhost ~]# cd /tmp/
[root@localhost tmp]# chmod +x NVIDIA-Linux-x86_64-430.99-grid.run
[root@localhost tmp]# ./NVIDIA-Linux-x86_64-430.99-grid.run --kernel-source-path=/usr/src/kernels/3.10.0-1127.19.1.el7.x86_64 -k $(uname -r)
Verifying archive integrity... OK
Uncompressing NVIDIA Accelerated Graphics Driver for Linux-x86_64 430.99.....................................................................................................................................................................................................................................................................................................................................................................................................................................................
[root@localhost ~]# nvidia-smi
Wed Aug 26 02:20:19 2020
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 430.99 Driver Version: 430.99 CUDA Version: 10.1 |
|-------------------------------+----------------------+----------------------+
| 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 P40 On | 00000000:13:00.0 Off | Off |
| N/A 28C P8 10W / 250W | 0MiB / 24451MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+
至此,透传成功!