Both Win10 and Win11 crash - AMD EPYC 7282-based system

[jaytee129]

I've created both win10 and win11 VMs on my new AMD EPYC 7282-based system with Supermicro H12SSL-CT MB and, with mostly out-of-the-box windows installations, the VMs crash hard after a while (no obvious pattern/timing to it).

Windows System Event Viewer gives these clues:

Dump file creation failed due to error during dump creation.

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

The previous system shutdown at xx:xx:xx PM on ‎2023-‎01-‎06 was unexpected.

Windows failed fast startup with error status 0xC000007B.

Focusing on Win 10 VM to start since Win10 more mature, here's my VM config:

agent: 1
balloon: 12288
bios: ovmf
boot: order=scsi0;ide2
cores: 20
cpu: EPYC-Rome
efidisk0: local-zfs:vm-201-disk-0,efitype=4m,pre-enrolled-keys=1,size=1M
hostpci0: 0000:c1:00,pcie=1
machine: pc-q35-7.1
memory: 16384
meta: creation-qemu=7.1.0,ctime=1672936369
name: GamerVM
net0: virtio=FE:E6:B7:C1:26:ED,bridge=vmbr0
net1: virtio=EE:84:A0:91:88:F9,bridge=vmbr1
numa: 0
ostype: win10
scsi0: local-zfs:vm-201-disk-1,discard=on,iothread=1,size=250G,ssd=1
scsihw: virtio-scsi-single
smbios1: uuid=50cd3213-74fe-412f-a65b-316cb02568e0
sockets: 1
vcpus: 16

Also tried cpu = host - same result

Don't want to use cpu=kvm64 because then I can't run VMWare Player inside the VM because it doesn't expose CPU virtualization capabilities

I have NVIDIA GE Force RTX 3060 passed through - could that be the problem?

Any suggestions/recommendations would be appreciated.

 

[jaytee129]

just to add/clarify that VMs crash even after I remove video card passthrough. are there BIOS settings that I should look to disable/tweak? Is it machine selection, i.e. should I use i440fx? is it disk controller selection - I have two PCIe nvme M.2 SSDs in zfs mirror? For the record, Proxmox is up to date as of today.

 

[jaytee129]

From a post elsewhere I see that disabling C-States in BIOS has fixed this kind of problem for others so I did that and so far so good. If anything changes, I'll post again. Otherwise all good.

I also learned that I can import a VMware VM into proxmox so no need to run VMWare Player inside a VM to run it. Very cool.

 

[slidermike]

Thank you for the updates @jaytee129
I am getting ready to do a similar configuration with my Milan build. This will help.
Do you happen to know which specific c-states setting?

I have an asrockrack mobo also. Not the same model but it is likely called the same in the bios, though in a sub-tab.
I see "global c-state control"