Hey,
Specs:
VMware ESXi 7.0.1
vSphere 7.0.1
Windows 10 Pro for WS 20H2 build 19042.631 VM
Tesla T4
I’ve installed the VIB on my hypervisor and it’s able to detect my graphics card.
I installed the 461.33 vGPU driver on VM and it’s successful. But when I reboot, I get a BSOD saying VIDEO TDR Failure for nvlddmkm.sys. I can’t figure out the solution for the problem. I’ve tried disabling all of the default VMware display drivers thinking there might be a conflict but that didn’t work.
HI, See more:
SYMPTOM OR ERRORS
When user tries to install vGPU drivers an error is seen “NVIDIA Installer cannot continue” with the details “The NVIDIA graphics driver is not compatible with this version of Windows. This graphics driver could not find compatible graphics hardware.”
ROOT CAUSE
It is very likely that the user has downloaded the wrong driver for their VM’s operating system. For example: this error will occur if a user is using a Win 8.1 32bits but attempting to install the 64-bit NVIDIA driver, cookie clicker!
SOLUTION
Ensure correct driver is used for the Operating system. The driver naming e.g. 362.13_grid_win8_win7_64bit_english should make it clear which version is installed.
RELEVANT PRODUCTS
NVIDIA GRID vGPU including K1, K2, M60, M6 GPUs
Windows Operating systems including 7.0, 8.0, 8.1, 10.0, server OSs
I have the correct driver. 461.33_grid_win10… and the installation runs successfully. It’s just when I reboot to fully complete the installation, the BSOD happens.
EDIT: 4/1/2021, so I really have gotten vDGA working. I was able to install the GRID vGPU driver and license it, etc and able to use it with my camera client. So it seems like the driver isn’t the problem, but maybe the hypervisor and VIB. I’m not sure. I believe I can use vDGA as a workaround for my use case.
Some other troubleshooting stuff: I’ve attempted doing vDGA and from what it looks like, everything is working. I don’t get a BSOD. I’m off network so I can’t really test it, but nvidia-smi recognizes the card.
So it seems like there’s something wrong with my implementation of vGPU. I’m not sure what though.
Log files attached.nvidia-bug-report.log (4.4 MB)
It sucks I can’t edit previous posts after a while…but I don’t think I can use vDGA as a workaround for my use case. I need multiple users to be able to use the gfx card at once, so I need to use vGPU. If anyone knows a solution, please please reach out. As you can see, I’ve been working on this for a week+ straight.
Hi ybaker613,
Sorry you are having issues. Can you check that you have set the Host Graphics Setting to “Shared Direct” - Virtual GPU Software User Guide :: NVIDIA Virtual GPU Software Documentation
D
I have verified that it’s set to Shared Direct. I can send a pic if needed.
Can you confirm that your T4 is installed in a compatible server for vGPU and VMWARE. A list of compatible servers are here: Virtual GPU Certified Servers | NVIDIA GRID
It’s not, but does it matter that much? vDGA and everything else for VMware works.
It matters because it is not clear if the issue is because of a incompatibility between your hardware; the T4 and the vGPU software. We work with OEMs/VMware etc to give buyers confidence that their hardware platform is compatible with vGPU and the T4 will run without error.