Linux Fixes Hosts Randomly Rebooting During Virtualization With Ryzen 7000/8000 CPUs

[WannaBeOCer]

Incorrect. It was a combination of a fabrication issue and silicon design issue(s) which Intel tried (and failed) to damage-control by rolling out a series of performance-reducing microcode patches.

There was a batch of contaminated wafers that are flawed. The majority of the issues is due to the microcode slowly roasting Intel's CPUs. Again it's the same issue as AMD's 7000X3D chips but more difficult to diagnose.

https://youtu.be/DznKg1IjVs0?feature=shared

[ssokolow]

Incorrect. It was a combination of a fabrication issue and silicon design issue(s) which Intel tried (and failed) to damage-control by rolling out a series of performance-reducing microcode patches.

*nod* See this video. The problem still occurred in datacenter machines where the server chipset didn't allow fancy tweaks and the used (default) configuration did everything but underclock the chips to prioritize reliability over performance.

[intelfx]

Two generations aren't flawed, it was a microcode issue

Incorrect. It was a combination of a fabrication issue and silicon design issue(s) which Intel tried (and failed) to damage-control by rolling out a series of performance-reducing microcode patches.

[Espionage724]

...
I agree their architectures are inefficient. What pushed me away from buying AMD is the USB issue that still plagues many AM4 users.

That bit me a bit with a Rift CV1 and X470 (on rear USB I couldn't push 3 sensors nor even just the HMD and luckily had a separate ASRock controller to push something to); a lower-spec'd Intel laptop handled the Rift with 3 sensors and random $10 eBay USB powered hub more stably surprisingly

[WannaBeOCer]

At least many models of Ryzen series have been very competitive products. Before Ryzen AMD had only shit hardware. Intel has f*cked the asses of their buyers several times without using lube. Two whole generations of recent chips are flawed and will eventually break down. Lots of unimpressive products. Totally f*cked up energy consumption for some generations of i7 CPUs. Also most of the 3rd to 7th generation models were totally unimpressive in terms of performance. 2600/2700k was good, the next good model was 8700k. Most users were also affected by SPECTRE/MELTDOWN. So much performance lost due to incompetence.

Two generations aren't flawed, it was a microcode issue. Unlike AMD who's microcode instantly fried the 7000X3D chips.

https://www.extremetech.com/computing/amd-to-limit-soc-voltages-with-final-update-to-chip-burning-saga

Intel slow roasted them. Making it difficult to diagnose the issue since they died much later than at launch. Having to go through a RMA does suck but hopefully they update their bios prior to installing their replacement.

I agree their architectures are inefficient. What pushed me away from buying AMD is the USB issue that still plagues many AM4 users. I went with the 12700K instead of the 5800X3D and since I game at 4K the CPU performance doesn't matter much unless playing a MMO.

[Espionage724]

Nobody pays me to counter every piece of bullshit I see on the Internet.

Fair enough, but it seems counter-productive to try to call out trolling without contributing something. I'm looking for discussion, what exactly are you doing?

[intelfx]

And yet you never have anything countering even upon request

Nobody pays me to counter every piece of bullshit I see on the Internet.

[Espionage724]

Lol, no, he doesn’t.

Judging by this guy’s recent comments across various topics, I’m practically convinced he is on some sort of a “trolling by cluelessness” contest.

And yet you never have anything countering even upon request

[yump]

Zen 4 aka Ryzen 7000/8000 does not support VMLOAD/VMSAVE. Never has. The hardware is there, because it's shared silicon with EPYC 4004, but it's not supposed to be enabled/advertised. It is, due to motherboard OEMs just enabling everything and shipping. Easy fix kernel side, means you don't need a firmware update.

In a more perfect world, we'd probably be able to dump out some supporting microcode/firmware enablement from a board that properly supports EPYC 4004 and enable VMLOAD/VMSAVE for everyone on Zen4, but it's really not worth the effort.

I was affected by this, replaced quite a bit of hardware trying to diagnose. Unlike some, I won't be making any melodramatic edicts about it.

I think it's safe to assume it's broken on 4004 too unless somebody comes along who actually has one and can demonstrate otherwise.

Mailing list message says nothing about it, and bugged hardware is bugged hardware.

[ssokolow]

This issue is about nested virtualization, i.e. VMSAVE/VMLOAD in the guest. You won't ever experience this issue unless you really go out of your way (and get unlucky with the BIOS/ucode).

*nod* I upgraded to a Ryzen 5 7600 at the beginning of the year, I do what should be fairly typical VM stuff, and it's been rock-solid for me. I imagine the only way a typical user might run into this is if they're trying to do something like virtualizing a Windows version new enough to have Virtualization-Based Security.