well, of course not installing the update will fix the problem, the thing is, those are cumulative updates, that means, no more updates
AMD : BSOD unsupported processor since Windows build 26100.4202+ ( update kb5060842 + its preview kb5058499 )
- Thread starter KciNicK
- Start date
The following settings resolved the "Unsupported Processor" issue:
args: -cpu 'host,arch_capabilities=off'
The following flags seem to be set when the CPU is host.
They were not present when using EPYC-v4 or physical installation.
args: -cpu 'host,arch_capabilities=off'
The following flags seem to be set when the CPU is host.
They were not present when using EPYC-v4 or physical installation.
Nice. That worked for me too.
Although I'm using the following custom cpu model rather than the args parameter:
Code:
cpu-model: host-fix
flags -arch-capabilities
reported-model hostHi, I enountered the same issue (UNSUPPORTED_PROCESSOR BSOD) with a Windows Server 2025 VM (not via Proxmox, but via german hoster Netcup that uses KVM-based virtual machines) after installing KB5060842 (Build 26100.4349). The host uses a AMD EPYC 7702P CPU.
Unfortunately I can't change the CPU type as a customer, and enabling "Virtual Machine Platform" didn't help in my case (as Netcup doesn't provide nested virtualization). However, I was able to work-around the bluescreen by reducing the CPU cores from 4 to 1 in the server control panel, after which Windows was able to boot successfully (but with reduced performance as only 1 CPU core is usable - as soon as I increase the CPU cores to 2 or more, the BSOD occurs again).
Edit: Netcup fixed the issue on 2025-06-17, though they don't say how. The CPU is still reported to the guest as "AMD EPYC 7702P", so maybe they also used the "arch_capabilities=off" setting.
Unfortunately I can't change the CPU type as a customer, and enabling "Virtual Machine Platform" didn't help in my case (as Netcup doesn't provide nested virtualization). However, I was able to work-around the bluescreen by reducing the CPU cores from 4 to 1 in the server control panel, after which Windows was able to boot successfully (but with reduced performance as only 1 CPU core is usable - as soon as I increase the CPU cores to 2 or more, the BSOD occurs again).
Edit: Netcup fixed the issue on 2025-06-17, though they don't say how. The CPU is still reported to the guest as "AMD EPYC 7702P", so maybe they also used the "arch_capabilities=off" setting.
Last edited:
I noticed that even Windows 11 IoT LTSC Enterprise is affected after the last patchday.
did you slipstream the last cumulative on that installation? anyway, i only see the bug on AMD processors
tried here, bosd during booting iso.
waiting official MS refreshed iso to test.
Last edited:
The cumulative patches have been installed.
I have confirmed that KB5063060 is installed.
This worked for me as well. I tried setting the CPU to Epyc-Rome-v4 and it would still blue screen like that on my 7800x3d.
What exactly does this argument do?
I believe this indicates, through CPUID, that a CPU supports the IA32_ARCH_CAPABILITIES Model-Specific-Register, indicating which side-channel mitigations are present.
See:
https://www.intel.com/content/www/u...cpuid-enumeration-and-architectural-msrs.html
What's curious is this seems to be an Intel CPU feature, yet it's enabled on AMD systems by QEMU/KVM by default on in host mode. There are various bugs/discussions on RedHat about this specific flag. On AMD, the mapping of side-channel mitigation in most mode may be incorrect.
It's conceivable changing this flag might impact side-channel mitigations within the guest. In my case, on a 5950X, by diffing
lscpu output with the CPU model changed from host to host-fix (see my prior post above) captured from a Linux Mint 22 adjacent guest VM, I didn't see any mitigation change. Your mileage may vary.Thanks for the detailed response.
I am actually not sure now that the the issue is resolved for me. I'm still getting fairly regular crashes right after booting the VM. I'm connected via Moonlight, though, so it disconnects before I can see the blue screen error code. I'm reverting to a previous VM save to confirm the issue is not related to GPU passthrough, as I'm having some weird issues on that front after moving from a X13SAE-F Intel mobo to the H13SAE-F that I'm on now.
Maybe this helps, i tried on all my AMD Servers.
Since others confirmed only AMD is affected, i didn't tested Intel Servers, but i can (have a lot of them)
Tested: W11 26100.4351 and W11 26200.5651
Ryzen 7 5800X -> Crash
Ryzen 9 9955HX -> Crash
EPYC 9374F -> Crash
Ryzen 7 PRO 8700GE -> Crash
I tested additionally a Native "W11 26100.4351 and W11 26200.5651" installation on Ryzen 9 9955HX, no issues!
I can test a Native Installation on EPYC 9374F too, but not the others sadly. If that helps, tell me.
So this seems to be a libvirt / qemu / kernel or proxmox issue only.
Cheers
Since others confirmed only AMD is affected, i didn't tested Intel Servers, but i can (have a lot of them)
Tested: W11 26100.4351 and W11 26200.5651
Ryzen 7 5800X -> Crash
Ryzen 9 9955HX -> Crash
EPYC 9374F -> Crash
Ryzen 7 PRO 8700GE -> Crash
args: -cpu 'host,arch_capabilities=off' fixes the issue.I tested additionally a Native "W11 26100.4351 and W11 26200.5651" installation on Ryzen 9 9955HX, no issues!
I can test a Native Installation on EPYC 9374F too, but not the others sadly. If that helps, tell me.
So this seems to be a libvirt / qemu / kernel or proxmox issue only.
Cheers
What do you mean by Native installation ?
Just tried with the official Microsoft updated Windows iso build 4349 (June 2025) : Cannot boot , bsod "UNSUPPORTED PROCESSOR"
Last edited:
it means without proxmox, direct install on the server