Guys quick help ! (BIOS : Fatal TLB error in CPU 0)

[devc1]

My stupid HP Z420 has a fatal problem, I entered bios settings and disabled and re-enabled ATS Support (By fault). I then rebooted my PC and when it came to the Windows login screen I touched the keyboard and sudden reboot, I taught that it was a tripple fault. When the PC booted it showed a message with an error beep "

Code: Select all

Fatal TLB error in CPU 0

Then the PC shutdown.
I powered it up multiple times after this, and the screen won't show, the cooler startup noise that it makes for a second is not present, like the CPU is stuck in an infinite loop.
I tried a CMOS reset, and nothing changing. When I kept it for a while the CPU keeped working until it heated and I started hearing the cooler accelerating...

Please I need help !


- I am not really sure that disabling it applied, but it had with it another 2 ATS features and they were disabled. I think they were enabled before.
- I didn't run my OS before this.
- Also previously, enabling secure boot would make an error beep and CMOS reset will make it work fine.

IS IT OVER ??

[nullplan]

Well, this sounds like the CPU has sustained damage. Something mechanical, since it only occurs when it heats up. I don't know if you can replace the CPU; a lot of laptop MBs have the thing soldered down. If that's the case in your system, your laptop just broke and you need to find a new one. Sorry, I wish I had better news.

[devc1]

Thanks for your reply, however I said that it showed the message and reboot without heating up. It is the first time it happened and now the PC is just stuck on an infinite loop without any message, warning or screen. Actually when I leaved it in its Power-On-Self-Test for a minute to test if the CPU is spinning for life, the CPU heated and the fan started working which proves what I said. I did remove the CMOS battery and now I am waiting some minutes before inserting it back.
Just for info, this is not the old laptop. This is my main PC (HP Z420 Workstation) which has a Xeon E51620 and a Quadro 2000 GPU, HDD AHCI drive. If it didn't work, I guess just solding its parts especially the CPU would be a huge refund since I got it for a medium or low price.

[devc1]

Bad news Removing the CMOS Battery, waiting for almost 20 minutes and plugging it back didn't solve the problem.

The only special thing that happened is : The PC booted for a second, then rebooted for another second then rebooted normally and stuck on the infinite loop with a low sounding somehow high frequency sound that I hear when windows is maybe communicating with the PCI (Moving the mouse, boot..). I think the ATS settings (Address Translation Services) did something to the PCI. And the CPU is actually looping on the PCI bus. Maybe,

[Octocontrabass]

It sounds like the CPU has failed. It's probably just a coincidence that the CPU failed right after you changed the settings.

[devc1]

The CPU never heated up from a long time, it has some little unaligned pins. I will check for this, I already knew that this PC will soon collapse because it had alot of problems before. Luckily, all my projects are in github and I can (or not) continue them on my very old laptop.

It is also a good occasion for me to lookup for an AVX512 PC, AVX is now unconsidered and everyone uses AVX2 instead.

[Octocontrabass]

It is also a good occasion for me to lookup for an AVX512 PC, AVX is now unconsidered and everyone uses AVX2 instead.

AVX is still used pretty extensively in code that needs to avoid throttling the CPU.

[eekee]

AVX is still used pretty extensively in code that needs to avoid throttling the CPU.

What does AVX2 do to slow down the CPU?

[devc1]

Please go open a new topic and talk about this.

In some cpus, the AVX(1/2/512) instruction set applies clock penalties. I guess they mean that the CPU slows down its clock when running an AVX instruction to avoid throttling

[devc1]

Octocontrabass, look. Isn't the CPU perfectly fine if the coolers work on heat ? The CPU must enter SMM mode to change their RPM !!

The ATS option changes did something to the bios startup, I am now trying an ME/AMT Flash override which I do not really know what it'll do. If it didn't work I will load my BIOS in a USB and try to flash it. I am sure there is something with the POST and it is related to the ATS options.

[devc1]

The ME/AMT Flash override Worked !

The PC boots up without problems, I have an overflow in my mind. The option I changed by fault was AMT Support and not Intel ATS. So this is why ME/AMT Hardware flash override worked.
Thanks.

Devc1,

[eekee]

In some cpus, the AVX(1/2/512) instruction set applies clock penalties.

Ah. Well, now you put it that way, I can imagine a few reasons why. Heat could be an issue, as you say. Using larger word sizes is just as bad for heat as increasing clock speeds. Or there's timing, which is crucial at an electrical level, and some operations are hard to make fast on large word sizes. So I guess I know all I need to about it. Sorry for the tangent.