Could someone help me to add a GPU GRID on a host?
Here is some information about the environment:
- Host: HPE ProLiant DL380 Gen10
- ESXi 7.0.2 (build 17867351)
- NVIDIA P6000 GPU
- NVIDIA-VMware_ESXi_7.0.2_Driver 470.63-1OEM.702.0.0.17630552 (VIB)
From vCenter, in the Hardware Graphics configuration session, the Active Type is appearing as Basic and available memory 0B.
But I already made the recommended settings and installed the VIB drivers. However, when I run some verification commands, some errors appear or nothing happens.
# esxcli graphics host get
Default Graphics Type: SharedPassthru
Shared Passthru Assignment Policy: Performance
# gpuvm - no output appears.
NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure the latest NVIDIA driver is installed and running.
But if I load the module manually with the command: esxcli system module load -m nvidia, the error changes to:
Failed to initialize NVML: Unknown Error
Xorg is also not running and as I read, it could be a BIOS or GPU firmware update, which apparently are OK.
I also read about having two graphics cards on the same server and that the GPU could not be addressed to be available to VMware. I also read something related with memory addressing limitation.
Anyway, I'm stuck on this... To make all these settings, I based myself on this link from VMware: Using GPUs with Virtual Machines on vSphere (part 3).
If you can just tell me what I'm missing, I'd appreciate it.
With the video card model in question, you can only perform Virtual Dedicated Graphics Acceleration (vDGA) where the card is passed directly to a VM and cannot be shared with others (what you are trying to configure is called Virtual Shared Pass-Through Graphics Acceleration but does not work with that type of card)
Here is the difference between the various types of use of video cards:
Deploying Hardware-Accelerated Graphics with VMware Horizon | VMware
Here are some indications of how to configure the vDGA
And here to see which cards are compatible: