Question GPU for desktop virtualization

Dec 27, 2023
1
0
10
Hello,

I'm completely new in the GPU world, but I have some plans and I would like to ask some help to make the right decision.

Currently I have a server with the following hardware components:
- SUPERMICRO X10SRi-F motherboard
- Intel Xeon E5-2620V4
- 3x8GB ECC DDR4 RAM
- 700W PSU
- 1 SSD and 6 HDD's

The main operating system is Ubuntu server and I currently use this server as a NAS. But the sever is a bit overrated for the this task, so I want to do some extra. The main idea is to make the server my home computer by using a virtual machine using KVM. The purpose is that it will be possible to access this virtual PC using RDP (Windows 11 as virtual machine) so that I can turn any device with a internet connection (VPN or internal network) into my personal computer, for example my phone, tablet, a Raspberry Pi,...

I already did some tests with this but I think I have some bottleneck of the graphical possibilities of my server. It works fine but when I start a video on YouTube in full screen mode, than it doesn't play the video in a fluent way, it looks its a video with 5FPS. I did this tests on my internal network so I don't think the network is the bottleneck. Is my conclusion correct that this is a GPU problem?

If this is correct what I want to do is add some GPU to the server and share this to the KVM machine, but I doesn't know which GPU is the best for this purpose. What I will do on this virtual machine:
- The standard things as Word and Excel
- Photo editing (Photoshop)
- Video editing
- 3D drawing (Inventor)

Currently I was looking for the MSI GeForce RTX 3060 VENTUS 2X 12G OC GPU. Is this a good choice for my purpose? And is it possible to share this GPU with more KVM's in the same hardware system, so that I can run more virtual machines next to each other and each system can use some part of the GPU?

If you need more information let me known!

Best regards,
Bram
 

Math Geek

Titan
Ambassador
in general a VM does not use the full capabilities of the gpu. it is only a very basic functionality as you noticed.

anything more than browsing and basic office work is too much for the virtual gpu in use. to use a full gpu you actually need to have a second gpu installed that you can pass through to the vm for its use.

this is not easy to do but it is possible. you'll need two different gpu sources. one for the server and a second one for the vm to use. once it is given to the vm it will be unavailable to any other vm or to the main system itself.