^Hi. I'm just a layman, but.. This is about VMs, I think. V86 and other CPU features should still work in Legacy mode on Ryzen (native boot).
Virtualizing them in x64 Long mode+Compatibility mode et cetera is a bit of a different story.
It adds things like Ring level -1, hardware-assisted virtualization et cetera.
https://en.wikipedia.org/wiki/X86-64#Operating_modes
That's the thing that old Ryzen messed up, among other things, maybe:
http://www.rcollins.org/ddj/Jan98/Jan98.html
http://www.rcollins.org/articles/vme1/VME_Overview.html
https://en.wikipedia.org/wiki/Virtual_8 ... ions_(VME)
Those "fixed" Ryzen CPUs maybe just dropped VME from CPU ID, they didn't fix anything.
If so, they just make software use an alternate code-path, a fallback.
Which in turn automagically seems to fix all the woes.
I think that's what early VM workarounds made use of, too.
They did overwrite CPU ID for the VM from the host side (explicitly remove VME flag in a VM's config file).
That got XP running inside the VM, at the time. Though it's notable that XP itself did the magic here, not AMD.
- Since I'm no Ryzen owner, I cannot test this, of course. So what I say is purely speculation and hypothetical. 😔
To double check, please use NSSI on Real-Mode DOS on a Ryzen PC and check CPU feature chart.
Let's please note that fallbacks for non-fully Pentium compatible CPUs were added into software way back in the 90s.
To get things running on Cyrix-like CPUs, so to say.
However, it's uncertain if those fallbacks from eons ago were ever properly tested. 🤷♂️
Their inclusion was more of a political decisions, maybe, than anything else.
All modern CPUs had VME, after all, including late 486 cores.
So a reliance upon their practical use is abnormal, strictly speaking.
Edited. Edited again. Formatting fixed.
"Time, it seems, doesn't flow. For some it's fast, for some it's slow.
In what to one race is no time at all, another race can rise and fall..." - The Minstrel
//My video channel//