During my testing of vGPU and GPU passthrough I ran into this an issue where I could not pass-through GPU’s with more than 8GB of VRAM.

In my case my GPU in my ESXi host was a Tesla K80 (24GB VRAM). Luckily, I stumbled upon a great article by Stephen Wagner who outlined this great. All this knowledge I’ve written below is from this post:

Stephen Wagner:

GPU or vGPU Passthrough with 16GB+ of video memory – The Tech Journal (stephenwagner.com)

NVIDIA Tesla K80

The VM Configuration:

First, the VM must be created with EFI under the boot options, and with secure boot enabled. Warning, this may cause your VM NOT to boot as it does mess with the VM Boot option, and the Operating System might not respect these new settings.

Next, we need to add the PCI Passthrough MIO advanced options to the VM itself.

The values you will be adding to your advanced settings of the VM are:

pciPassthru.use64bitMMIO=”TRUE”

pciPassthru.64bitMMIOSizeGB=32

Then, make sure your PCI-E device is passed through to the VM, and save your configuration.

Then the VM should be good to start, and should boot with the PCI-E device attached correctly.

If you have any questions, comments, or need help comment below!