PC Processors

[username007]

I am facing a very strange issue. My setup is as follows

7950x (no overclocking)

Asus x670e Hero Motherboard

2x32GB Kingston KF556C40BBK2-64 DDR5 RAM (tried both DEFAULT and EXPO1 Profiles)

1350W Thermaltake ATX3.0 PSU

MSI 4090

I have ran several stress testing tools (MEMTEST, PRIME95, OCCT) for many hours and I am unable to find any errors.

The issue occurs when I run a virtual machine for Android development work

Windows Server 2022 Standard Host OS

VMWare Workstation Pro 17

Windows Server 2019 guest OS

I can work in Android studio within the guest OS without any issues for long periods of time, but when I start an Android emulator within Android Studio for testing my application, everything works for some time, but then my PC reboots/crashes.

The virtual machine has been assigned 4 cores and 16GB RAM. I have not tested. I haven't tried running Android Studio in my HOST OS yet. I don't see any issues while running stress testing tools for long periods of time. I have not enabled any PBO settings in BIOS and have updated my bios to the latest version available

[elstaci]

In case you don't get to much action here:

Not sure if you should post this thread at AMD Developer's Forum or AMD Server Guru.

You are using Server OS but also using VMware and Android.

Anyways here is the link for AMD Developer's Start Here: https://community.amd.com/t5/newcomers-start-here/bd-p/newcomer-forum

The link for AMD Server Gurus: https://community.amd.com/t5/server-gurus/ct-p/amd-server-gurus

[atmosphere]

So you get a hard crash, e.g hang or reboot right? I presume you mean the host crashes

Within vmware do you use any passthough devices? If so update the chipset drivers. The passthough option captures the port from the OS, so a driver bug could cause a hard crash, as well as a faulty USB cable.

Please try the chipset driver (for improved usb support)  

Also MS has a history of updating things in favor of HyperV. So windows update and also check for VMWare WS update..

[username007]

Yes, hard crash. The screen goes blank and the PC restarts. There are no passthrough devices. I just tried virtualbox, and it crashed again. The difference was this time it crashed while installing Windows within Virtualbox VM. So the crashes seem to happen when doing something with VMWARE/VIRTUALBOX. I will try updating the chipset drivers

[misterj]

username007. Please look in "C:\Windows\Minidump" and if not empty please compress it a put it were I can DL it. Also look in the Event Viewer and post a couple of the errors occurring, Basic view. What OS are you using on the HW. Why are not using Hyper-V? Do you have SVM enabled in BIOS? Please post a relevant screenshot (don't know what to ask for yet). Make sure all your AMD drivers come from here. Also remove any applications from your MB vendor and make sure you are on the latest BIOS. Thanks and enjoy, John.

[username007]

The only event logged in event Viewer is

The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.

And "C:\Windows\Minidump" is empty. But today (while installing Windows under Virtualbox) there was a file in "C:\Windows\Minidump"

I have uploaded it here

Minidump 

I already had a license for VMWare so didn't look into HyperV. SVM is enabled in BIOS

[misterj]

username007, this is DRIVER IRQL NOT LESS OR EQUAL. Google says make sure your drivers are up to date. Make sure all your AMD drivers come from here. Did you do a fresh install of Widows on this processor? I used a free version of VMWare and liked it but I like Hyper-V much better. What version of Windows are you using? I have no experience with all these VMs on W10 and suggest you pick one and stick to it until we determine what is going on. I would also suggest you uninstall any applications from your MB vendor and only get your BIOS (Latest) from them. Thanks and enjoy, John.

[username007]

I am using Windows Server 2022 (fresh install). I just updated all drivers and will currently I am testing if that made a difference. Thank you

[atmosphere]

So thats a crash allright. So one or more drivers arent handeling IRQ as it should, or triggers a NMI. 

But in case of a hard crash, it does not show the ":-("-screen?! Hardcase are mostly kernel-mode drivers, kernel, firmware or hardware faults. (Besides an possible bit-flip that is.) 

You have latest BIOS , Drivers, windows-updates* so that should play nice. 

(* => i've had an issue myself with server2019 that on two VM's i had both with internet acces for updates during setups. If i opted to search for update, the setup failed or crashed @1st boot. The other one needed the updates. 

For scope: about 20 same OS servers on that cluster installed just fine, running great.

Disable secure boot, and driver signing.. but only when using official media from MS-site. Use BIOS-mode for the VM at legacy/BIOS or EUFI.

On the host, did you enable the option to expose visualisation to vm's. Also there is an option in server 2022 to enable virtualisation for about everything within windows. (Virt. Based Security)

Also in BIOS/EUFI on host, what are options for io-srv iommu and above 4G-decoding?

And luck with chipset drivers on host btw? AMD dropped new drivers recently.

[username007]

Are you running Windows server on non-server hardware?.

I am running Windows Server 2022 host on x670e Chipset and 7950x and since there are no drivers available from AMD with Windows Server support I have installed Windows 10 drivers on Windows Server 2022 for the chipset and all other hardware. I went back to Windows 10 and ran the same VM on the same version of VMWare Workstation on Windows 10 host, and the problem doesn't happen on Windows 10, no crashes. So it seems like I cannot run Windows Server 2022 with Windows 10 drivers and expect everything to work

Though every thing is rock stable on Windows Server 2022 this particular use case (nested virtualization, runiing android emulator inside windows VM) causes crashes 

[nonametoday]

I'd test the RAM since RAM can cause really odd behaviour. So i would advice you to just run Memtest86 and see if you pass it. Memtest is really good at spotting problems related to RAM so you might give it a try.

[username007 · Accepted solution]

For anyone else running a similar setup and runs into same problem

I have been running Windows Server 2022 host on x670e Chipset and 7950x and since there are no drivers available from AMD for the chipset with Windows Server support I have installed Windows 10 drivers on Windows Server 2022 for the chipset and all other hardware. For the most part everything works fine.

I ran into issues when I was doing Android development using Android Studio in a Windows server VM running under VMWare Workstation Pro version 17 on my windows Server 2022 host. The VM works without issues but when I start an Android Virtual device to test my app, after some time my system (host) crashes (screen goes blank and the system reboots without any BSOD or minidump). I have ran all sorts of stress tests for 8+ hours (MEMTEST, OCCT, TESTMEM, GPU Stress test) but none of them found any issues

I finally went back to Windows 10 and ran the same VM on the same version of VMWare Workstation on Windows 10 host, and the problem doesn't happen on Windows 10, no crashes. So it seems like I cannot run Windows Server 2022 with Windows 10 drivers and expect everything to work