Issue with installing NVIDIA Graphics Driver

Graphics / Linux Linux
1

Hi,

I have HP server Dl360 with GP102GL [Tesla P40] installed.

the server OS is VMware 6.7 and on top I have a VM Centos 7.7.1908.

The card is configured as passthrough.

When running the run file (NVIDIA-Linux-x86_64-450.80.02.run) I get the error unable to load the kernel module ‘nvidia.ko’.
nvidia-installer.log (27.6 KB)

rpm -qa | grep -i kernel

kernel-3.10.0-1062.el7.x86_64
kernel-tools-3.10.0-1062.el7.x86_64
kernel-tools-libs-3.10.0-1062.el7.x86_64
kernel-devel-3.10.0-1062.el7.x86_64
kernel-header-3.10.0-1062.el7.x86_64

I have disabled nouveau with the following commands:

grubby ––update-kernel=ALL ––args=“rd.driver.blacklist=nouveau nouveau.modeset=0”
mv /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r).img.bak
echo “blacklist nouveau” > /etc/modprobe.d/nouveau-blacklist.conf
dracut /boot/initramfs-$(uname -r).img $(uname -r)

reboot

Please advise,
Aviram

2 
[   89.199760] NVRM: This PCI I/O region assigned to your NVIDIA device is invalid:
NVRM: BAR1 is 0M @ 0x0 (PCI:0000:13:00.0)

Please make sure 64bit BARs ar enabled in bios and the OS is installed using EFI-boot.

3

My os is installed using BIOS and not EFI. will that be an issue?

4

Yes, both requirements have to be met, the Tesla won’t work with bios boot. Please create a new vm and reinstall using efi boot.

5

Additionally, you might have to set the address space of the vm:

 pciPassthru.use64bitMMIO=“TRUE”
 pciPassthru.64bitMMIOSizeGB=“64”

https://blogs.vmware.com/apps/2018/09/using-gpus-with-virtual-machines-on-vsphere-part-2-vmdirectpath-i-o.html