A100 not recognized in VM

Hey everyone,

we have two brand new Dell PowerEdge R7525 with one A100 each. We want to run VMware vSphere ESXi 7.0 Update 2 with MIG. We enabled SR-IOV in the BIOS, changed default graphics type to shared direct and installed NVIDIA-VMware_ESXi_7.0_Host_Driver 460.107-1OEM.700.0.0.15525992.
But after enabling MIG we aren’t able to get the GPU running inside the VM.
If we create a GPU and Compute instance using nvidia-smi on the ESXi and attach the GPU using the corresponding profil in the VM settings, the VM fails to start with the error message: " vGPU ‘grid_a100-1-5c’ failed to create MIG GPU instance. Failed to start the virtual machine. Module DevicePowerOn power on failed. Could not initialize plugin ‘libnvidia-vgx.so’ for vGPU ‘grid_a100-1-5c’."

If we didn’t create the GPU and compute instance, we are able to boot the VM and see new GPU and Compute Instances on the ESXi, but if we try to install the GPU driver, the installer fails with the error message, that no compatible graphic card was found.

What have we missed or are we doing wrong?

1 Like