Introduction to NVIDIA Virtual GPU - Part 2 - vGPU Sizing and Common Mistakes

NVIDIA Solutions Architect, Lee Bushen, will discuss all things NVIDIA Virtual GPU (vGPU) in this three part video series.
Part 2 of 3
Lee will discuss GPU deployment. In this video we cover some of the common mistakes that people make when sizing their vGPU-enabled desktops. We also cover some recommended starting points and some tools to help you get the right vGPU Profile. We'll finish by talking about how you can demonstrate to others how much better the density and user experience is with vGPU-enabled VDI desktops.
####################
CHAPTERS IN THIS VIDEO
####################
0:00 - Introduction
01:06 - Common Mistakes
13:33 - Sample Configurations
17:10 - Sizing Tools
21:29 - Convincing the Boss
24:40 - Summary
Learn more about how to size #NVIDIAvGPU here
- Slide Deck www.slideshare.net/LeeBushen/...
- Useful web links to demonstrate vGPU advantage drive.google.com/file/d/1K33M...
- NVIDIA Virtual PC Sizing Guide nvda.ws/3nffgYz
- NVIDIA Virtual GPU Positioning Guide
www.nvidia.com/content/dam/en...
#NVIDIAvPC #NVIDIARTX #Virtualization

[@patrickbader5742]

Thanks, useful information.
After building a VDI Cloud for CAD I learned all the points in this video real life. Maybe it would have helped to have access to this ressource 2 years ago ;-)
In 2:00 you mentioned AutoCAD as an example for a higher needed vGPU Profile. From my experience AUtoCAD is in 95% of the time a singe Core CPU application. Adding a vGPU never really added benefits, sometimes it even created more problems which led to disable hardware acceleration. I think this is not a issue of vGPU but AutoDesk.
Designing and sizing a vGPU solution around the application is probably the key takeaway.

[@rod85y]

Is there no way for passing through from your PC to a hyper-V virtual machine running on it?

[@Hamkaastosti1]

Yes you need to use DDA, this will however not use vGpu profiles. You are passing trough the whole card, or in case of a M10 with 4 gpu cores you can assign this card to max 4 VM's