1

Our built is a MSI B550 A PRO with a Ryzen7 5800X and two 16GB g.skill F4-3200C16-16GVK in slots a2+b2 and an RTX3060. The PSU and Cooler are be.quiet and indeed very quiet and powerful. Heat is not a problem in this system. The CPU in "office" work is 40...50 deg only in hefty gaming and performance testing it approaches the 90deg, but never exceeds (frequency is limited in this case).

This system works since two years with only occasional glitches - in hindsight very likely indicating something is not ok.

The mainboard bios is the most recent one. All firmware is up-to-date.

Yesterday I ran a "stress" test in linux with "s-tui" for >3hours without problems. I also, partially at the same time ran "gpu_burn" for 2 hours without any problems or Errors. The CPU was always maxed out, all cores around 4.5GHZ and the temperature at 90 deg. The GPU never go warmer than 55 deg.

While this is very nice, it became very untypical these days:

During the last days the systems becomes more and more unusable, crashing often after minutes mostly without error messages and in all kind of random situations. At low-load or high-load, or during boot, anytime.

When I boot into memtest86+, it can run for hours without any errors. The CPU gets to 40..50deg in this case and one full pass takes roughly an hour. I even played with Memory settings in the bios, the memory speed varies a bit, but there are never any errors.

I also ran extensive testing on the two M2 SDD drives, and they are also without any errors.

We normally run Linux for office and gaming work (ubuntu or arch). But there is also a plain Windows. Now, none of this really even boots up any more, or runs stable for more than minutes. Mostly the system just reboots, sometimes it also freezes. In Minecraft is threw many "Error 6"s, which it never did before at all.

Error messages grabbed with my phone from the screen in rare occasions are:

during a failed ubuntu boot:

mce: [Hardware Error]: CPU 9: Machine Check:0 Bank 1: baa0000006
mce: [Hardware Error]: TSC 0 MISC d012000100000000
mce: [Hardware Error]: PROCESSOR 2:a20f10 TIME 1698596062 SOCKET 0 APIC 3 microcode a20102b
NVRM: loading NVIDIA UNIX x86_64 Kernel Module 545.23.06 

while running some ubuntu shell code:

[Hardware Error]: Corrected error, no action required 
[Hardware Error]: CPU:9 (19:21:0) MC1_STATUS[-|CE|MiscV|AddrV|-|-|SyndV|-|-|-]: 0x9c20000006010859
[Hardware Error]: Error Addr: 0x00000000f7c1b8d0 
[Hardware Error]: IPID: 0x000100b000000000, Syndrome: 0x000000005a008323
[Hardware Error]: Instruction Fetch Unit Ext. Error Code: 1, IC Microtag or Full Tag Multi-hit Error.
[Hardware Error]:  cache level: L1, mem/io: IO, mem-tx: IRD, part-proc: SRC (no timeout)

And also in /var/log/kern.log there are often many such MCE status dumps of all CPU cores (I don't have access to the log right now).

A Windows boot ended with

Stillstandcode: IRQL NOT LESS OR EQUAL

A new one: After 1:42h of memtest86+, in pass three, with 0 memory errors

Unexpected interrupt on CPU1

Type: Invalid Op
IP: 10b974
CS: 10
Flag: 10002
Code: 0
DS: 18
ES: 18
SS: 18
rax: 75c08445
rbx: 0
rcx: 1
rdx: 0
rdi: 4b
rsi: 1151ce
rbp: 116019
rsp: 12b220

To me this look very much like there is something wrong with the CPU. I just never had a CPU failing on me in many decades. In particular after working for a long time and being treated nicely (no overclocking, no overheating, etc.).

Can I do any more diagnostics on my system to understand better what really broke? What is the best strategy to get back a working system? New CPU? New CPU+Motherboard? First checking that the RTX3060 is really ok - but I have no indications that there is anything wrong with it?

4
  • Get the Manufacturer's hardware test diagnostics and test the hardware
    – John
    Oct 29, 2023 at 17:59
  • Interesting idea, but I am not aware of any such tools from AMD or MSI. I mean, there are all kind of "monitoring", but this I all have, and then there is generic stress testing - this I did (but it is unspecific). Could you point to something as an example? Oct 29, 2023 at 19:12
  • You would need to search for the mainboard model number to see what is there.
    – John
    Oct 29, 2023 at 19:13
  • Ok. There are only fancy monitoring tools offered by MSI, no diagnostics. I can see all the stats of the board (rpms, v, Temps, etc) and all are OK. Oct 30, 2023 at 5:13

1 Answer 1

0

To let you know, indeed it was the CPU only that was defective.

However, the only way to find this out is the hard way. Replace indivudal parts, search for a test system, keep running tons of stress tests and diagnostics.

A main hint for us was a crash in memtest86+ running it for many passes. There never was a hint of a memory problem, the crash clearly pointed to cpu (or mainboard). We swapped CPUs and mainboards to be really sure...

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .