Windows 10 Gaming VM with Nvidia

[Mihael]

 

Hello everybody, I am slowly despairing...

 

After I had problems getting the VM running with an AMD GPU (5700), I was able to swap it for a Nvidia (gigabyte geforce rtx 2070 gaming oc).

 

 

At the beginning, the VM was working properly and I was able to run the VM with GPU passthrough too and I tested a game and everything worked super smooth.

 

 

Unfortunately I did a force shutdown and I could not start the VM anymore and the log quickly went up to 100%. Extending the log from 128mb to 512mb does not help, as it quickly gets full after another VM startup. Only a server reboot helps to clear the log files.

 

 

Since the graphics card has an USB-C port, I've added the usb-c & serial bus controller ID to the syslinux config as recommended in a video.

As you can see in the picture, these 2 also appear in "Other PCI Devices"

 

10de:lada

10de:ladb

 

 

 

Although I can attach the graphics card and the sound of the VM, but I can not start it anymore. The 4 attached CPU cores and log is running at 100%

 

 

Does anyone have any idea?

 

 

In the second PCI-E slot, I have a P2000 passthrough to Plex docker and this works fine since weeks.

 

Thanks!

 

[meep]

If you pass through the USB and Serial controllers to the VM (check the ' other PCI devices' boxes), will it boot?

If you remove the USB and Serial controllers from your syslinux, will the VM boot?

 

Is there anything in the log file that's repeating over and over that causes it to fill so quickly?

 

[Mihael]

14 minutes ago, meep said:

If you pass through the USB and Serial controllers to the VM (check the ' other PCI devices' boxes), will it boot?

If you remove the USB and Serial controllers from your syslinux, will the VM boot?

 

Is there anything in the log file that's repeating over and over that causes it to fill so quickly?

 

"If you remove the USB and Serial controllers from your syslinux, will the VM boot?"

 

It didn't worked before, and I found this little tutorial

 

 

After adding those two "devices" to the syslinux it worked. But after an unfortunate "force shutdown" it doesn't work anymore.

I will try to remove and start without that or to check the other PCI devices boxes and let you know.

 

Thanks.

 

[Mihael]

I added the diagnostic file here->

diagnostics-20191208-2017.zip

[meep]

Something else to try. Rather than rebooting the server, shut it down completely, remove the power cable for 2 minutes, then boot it back up.

 

in an old system, I had to do this to force a gpu to reset. Only a complete power removal would do it.

[JonathanM]

29 minutes ago, meep said:

remove the power cable for 2 minutes

On most systems pushing the power button while the power cable is unplugged will accomplish the same thing as waiting a few minutes.

[Mihael]

22 minutes ago, meep said:

Something else to try. Rather than rebooting the server, shut it down completely, remove the power cable for 2 minutes, then boot it back up.

 

in an old system, I had to do this to force a gpu to reset. Only a complete power removal would do it.

I shut it down completely and removed the power cable for 5 minutes... no help. As soon as I start the VM, log goes to 100% and can't connect the VM via remote desktop and no signal to the screen.

If I remove those syslinux entries and reboot the server, I get this message ->

 

I thought that this GPU reset is only necessary with AMD GPUs. 

 

Do I need a third GPU for Unraid? 

 

I see in the older Unraid tutorials that they add a third GPU. I use my second GPU for the Plex Docker and this works without any problems.

 

The VM works without any troubles if I start the VM with VNC instead of GPU passthrough...

[meep]

So it looks like you need those syslinux entries.

 

i cannot look at that .zip file right now as I’m on a tablet. Have you tried swapping your Gus between VMs?

 

is this problem gpu also your unraid boot gpu? Have you tried including a rom file?

 

 

[Mihael]

25 minutes ago, meep said:

So it looks like you need those syslinux entries.

 

i cannot look at that .zip file right now as I’m on a tablet. Have you tried swapping your Gus between VMs?

 

is this problem gpu also your unraid boot gpu? Have you tried including a rom file?

 

 

"So it looks like you need those syslinux entries."

It's because the USB-C connector on the GPU I guess...

 

"Have you tried swapping your Gus between VMs?"

I have just the one Windows Gaming VM, but I can try to swap the GPUs between the Plex Docker and the Windows VM.

 

"is this problem gpu also your unraid boot gpu?"

I guess so. I can see how unraid boots on my screen. I have a Displayport connector on my motherboard, but I don't know how to make that one as the Unraid GPU. I have to mention that I have a AMD Ryzen 3700x, as far as I know it doesn't have a GPU onboard, but the Asus mainboard has a Displayport connector.

 

"Have you tried including a rom file?"

No, but I will try to do that tomorrow after work.

 

Thanks a lot for your time!

 

[testdasi]

Surprised no body asked for:

• Tools -> Diagnostics -> attach zip file
  • Preferably do this AFTER you start your VM (and fail and/or syslog shoots to 100%).
• Attach your VM xml (if copy-paste, use the code function - the </> button next the smiley button)

Syslog shooting to 100% suggests you have an actual problem that is spamming your log.

 

[Ver7o]

Have u tried making a new VM with the same setting to test if that one would throw the same errors?

[bastl]

@MihaelInterestingly it looks like the USB controller of some 20xx series cards is causing the same reset issue like some AMD cards show for a while now. The usual fix is to not passthrough the controller and not connect anything to it. I remember 2 users now which had issues with it and this fixed it for them but I can't remember if they stubbed the devices or not. If you append them to the syslinux try to add all devices from the card and not only the 2 controllers to prevent unraid to initialize the whole card.

[Mihael]

I've been pretty busy for the past few days and I was also abroad.  I wanted to thank you for the many suggestions and the help.
 

The Unraid server is now running smoothly since Unraid 6.8.1 (NVIDIA BUILD). I am now also satisfied with the RTX 2070 gpu. A lockdown rarely happens, the last lockdown happened with Unraid 6.8.1 (NVIDIA BUILD)

 

At the beginning I had a lot of problems with my UNRAID Server and the main reason was that a RAM module was broken. Although all 4 modules were new, one was broken. I could only find out after a MEM TEST. The Mobo bios was also upgraded to the latest version.

 

 

The server takes over several tasks at the same time without issues. I can play FarCry 5 with max settings while 4 Plex transcoding are running. It could handle more transcoding, but I haven't tested it.

At high load, the inside of the case gets around 36° celsius warm. The hard drives have more or less the same temperature.

 

Below I have listed my setup if anyone is interested:

 

- Plex Docker with GPU Passthrough (Nvidia P2000) 

- Tautulli Docker

- SabNZB Docker

- Radarr Docker

- Sonarr Docker

- Deluge Docker

- Make MKV

- Handbrake

- Ombi

- Upstatsboard

- Filebot

 

- Windows 10 (1909) VM with GPU Passthrough (Gigabyte GeForce RTX 2070 SUPER Windforce OC) and one USB port is assigned for mice and keyboard - just for Gaming

- Windows 19 (1909) VM without GPU Passthrough - Random work

- Mac OS Catalina VM - Random Tests

 

Hardware:

 

CPU:              AMD Ryzen 7 3700X 8-Core @ 3800 MHz

CPU Cooler:   Corsair H100i
Mobo:            ASUS ROG STRIX X570-E (Bios 1405)

SSD:              2 x 2TB PNY XLR8 CS3030 for Unraid Cache1 & Cache2
RAM:              4 x 16GB Corsair Dominator Platinum 3200 MHz

GPU 1:            NVIDIA Quadro P2000

GPU 2:           Gigabyte GeForce RTX 2070 SUPER Windforce OC 8G
Storage:         9 x 10TB Seagate IronWolf & 1 x 10TB Toshiba X300

PSU:              Seasonice Prime Ultra Titanium 1000W

Case:             Fractal Define R6

 

 

[Vilse(swe)]

I think i might know the problem isent unraid its hardware / BIOS.

 

Display Port cabel have a bad habit of making BOOT problems.

The system draw power true dp cabel and shorcut bios from booting and it only apers after shutdown/reboot.

On real cold bot it offen works when power is total depleeted.

It tock me 4 days to find out way my pc dident bot untill i removed PD cabel.

I contacted my Mainbord manufactory and thy said i had to flach bios ovh try changing boot mode.

In the end i just change to HDMI problem solved