Unable to configure A100 on a vSphere 7.0.3

Virtual GPU Forums More vGPU Forums NVIDIA Virtual GPU Technology
1

Hi,
tried to configure A100 (GA100 [A100 PCIe 80GB]) on ESXi 7.0.3 but nvidia-smi fails:

[root@xxx:~] nvidia-smi
NVIDIA-SMI has failed because it couldn’t communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.

  • Latest driver NVIDIA_bootbank_NVIDIA-VMware_ESXi_7.0.2_Driver_510.85.03-1OEM.702.0.0.17630552 installed:

[root@xxx:~] esxcli software vib install -v /directory/NVIDIA_bootbank_NVIDIA VMware_ESXi_7.0.2_Driver_510.85.03-1OEM.702.0.0
.17630552_77c428eb-3731-4fbf-95da-394c4e086433.vib
Installation Result
Message: Operation finished successfully.
Reboot Required: false
VIBs Installed: NVIDIA_bootbank_NVIDIA-VMware_ESXi_7.0.2_Driver_510.85.03-1OEM.702.0.0.17630552
VIBs Removed:
VIBs Skipped:

  • MMIO configured for 14TB
  • Integrated Video Controller disabled
  • xorg failed to start
  • Host Graphics Settings
    Default graphics type - Shared Direct
    Shared passthrough GPU assignment policy - Group VMs on GPU until full (GPU consolidation)

any idea?

2

Did you manage to get it running, think I´m in the same boat as you are.

3

Hi @weidi1 yes I did :) A100 needs EnterpriseAI license… i explained it here How to configure vSphere 7 with Multi-Instance GPUs (MIG) or Time-Sliced Profiles on the NVIDIA A100 - VIRTUALINCA

1 Like
4

Hi,

where did you get the host driver from? Be aware that you need NVAIE license for this use case and the matching host driver from the NVAIE package to contain the right device IDs for this GPU

Edit: vesic was a bit faster with his response :)

1 Like
5

I was hoping that the usual vGPU driver would magically work but was tought better ;)