Weird new issue

deviation · Sunday at 09:04

So this occured a few times at the beginning of last week, and then hasn't happened since, then it occured on Thursday night once, and then last night twice...
The OS just randomly restarts the PC, when you're barely doing anything, it's happened when just sat on the forum, sat idle, but never when gaming?
Last night it did it after hours of use, then again within 10 minutes?

I haven't done anything to the PC other than recently tidy up the cable management... But I did check and nothing is loose?

Event Viewer has this to say:

Event 41, Kernel-Power

The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly. (70368744177664)

Could my ram have gone bad? I tried to test it and it caused one of the restarts, usually it would halt on error after hours of loops. So I'm unsure if it is the ram in this case?

EDIT:
It's just done it again, this time whilst running Memtest64 on a loop with nothing else running, same Event Viewer error!

This occurs in both Windows and Linux, so it's not a software/OS issue.

Any ideas?

Thanks!

Firegod · Sunday at 10:48

Have you tried reseating the RAM? Triple check power connectors at both ends as well (mobo and psu). If it is crashing when running the memory test that does suggest faulty RAM. Have you tried again with x1 DIMM or stock timings if using XMP or EXPO?

deviation · Sunday at 10:52

Firegod said:
Have you tried reseating the RAM? Triple check power connectors at both ends as well (mobo and psu). If it is crashing when running the memory test that does suggest faulty RAM. Have you tried again with x1 DIMM or stock timings if using XMP or EXPO?

I haven't, but I will try that when I test 1 stick at a time with the Memtestx86 bootable USB - I'm just currently waiting 2 hours for Windows to 'Fix' my drive, I stupidly decided to do a chkdsk scan :cry:

This problem occurs in Linux as well, so it is definitely hardware related and not a problem with software/OS.

I'm going to disable XMP and try that next, when I got the ram I did do a 8, 12, 24 hour Memtest and it passed with both XMP profiles, so it would be strange if it's randomly decided to not like it anymore, but I agree it sounds like the ram, it usually is, but I've never seen a restart when testing it before, usually it halts on errors, when I've tested faultly sticks.
I have checked all power connectors 3 times haha, twice this morning and once last night

Quartz · Sunday at 11:38

deviation said:
I haven't done anything to the PC other than recently tidy up the cable management... But I did check and nothing is loose?

You might want to go back and check that all the lockable connections are tight and locked - at both ends.

deviation · Sunday at 12:30

Quartz said:
You might want to go back and check that all the lockable connections are tight and locked - at both ends.

Thanks, yeah I made sure. Will check again when I swap out my spare sticks of ram.

If it's going to be anything, I just want it to be the ram, if it has to be something

deviation · Sunday at 12:44

I've decided before I pull the existing ram to Memtest86 it with the xmp disabled by manually specifying the speed, I've seen that work before on ram that is higher than the supported speed, this way I will know if it's just XMP playing games, despite it working before fine when tested :confused:

My moneys on the ram has become faulty.

deviation · Sunday at 16:18

Opinions @Quartz @Firegod

do you use Memtest86+ the open source free program https://memtest.org
or the free or paid versions of memtest86 by PassMark https://www.memtest86.com

In the past I think it was the open source first option that found when I had errors before? But I have definitely used both.

I've just finished doing 4 passes with memtest86 by PassMark, and passed with XMP disabled and manually setting the MHz for the ram...
I'm going to do another 4 passes with XMP turned back on to see if I can get it to show any errors.

Which do you find is the more effective program?

Mr Mcgoo · Sunday at 18:42

memtest86+ is better as it has more tests. It's also completely free. last update was around 2022 which gives full support for ddr5 and will be updated again once ddr6 comes around.

deviation · Sunday at 19:04

Mr Mcgoo said:
memtest86+ is better as it has more tests. It's also completely free. last update was around 2022 which gives full support for ddr5 and will be updated again once ddr6 comes around.

Cheers mate, I was hoping that was the case, the paid for one looks a bit of a con, charging you just to unlock looped runs etc.
I'm pretty sure what I used in the past was the opensource free one

qbazdz · Sunday at 23:43

What's the PSU?

aaronyuri · 2025-07-21T01:31:48+0100

deviation said:
Opinions @Quartz @Firegod

do you use Memtest86+ the open source free program https://memtest.org
or the free or paid versions of memtest86 by PassMark https://www.memtest86.com

In the past I think it was the open source first option that found when I had errors before? But I have definitely used both.

I've just finished doing 4 passes with memtest86 by PassMark, and passed with XMP disabled and manually setting the MHz for the ram...
I'm going to do another 4 passes with XMP turned back on to see if I can get it to show any errors.

Which do you find is the more effective program?

Memtest86+ all day long!

Bear in mind that if you're manually setting DRAM frequency and leaving the timings on auto, the mobo will set really lenient timings; they will be nowhere near as tight as the XMP profile. For a true baseline test, it would be better to load optimised defaults and perform stability testing that way.

However, since it's been working for a while, test with XMP enabled and reseat the DIMMs just to cover that off. Memtest86+ tests 7 and 8 (random, IIRC) are usually good at catching intermittent memory faults.

Can you post your full rig specs?

deviation · 2025-07-21T07:30:25+0100

qbazdz said:
What's the PSU?

Corsair SF750 80+ Platinum.

aaronyuri said:
Memtest86+ all day long!

Bear in mind that if you're manually setting DRAM frequency and leaving the timings on auto, the mobo will set really lenient timings; they will be nowhere near as tight as the XMP profile. For a true baseline test, it would be better to load optimised defaults and perform stability testing that way.

However, since it's been working for a while, test with XMP enabled and reseat the DIMMs just to cover that off. Memtest86+ tests 7 and 8 (random, IIRC) are usually good at catching intermittent memory faults.

Can you post your full rig specs?

It's just done 12 passes for 10 hours 14 minutes with Memtest86+

completely error free with manual DDR MHz set, so that's promising, I've set the MHz back to auto and enabled XMP profile 1 again, when I got this ram I did 8-12 hours on both profile 1 and 2, so I know it did work fine with no errors.
I have reseated the ram and unplugged and plugged in both sides of all the plugs, so we shall see how the next 10-12 hours of Memtest86+ testing goes

I'm hoping it's fixed now and somehow I must have knocked something when messing around in there redoing my cable management recently, but I did check it 3 times prior to redoing it again yesterday

Either way if it's fixed and I don't have to buy new ram or PSU, then I'm happy :cry:

I just hope real world it then behaves, as originally it didn't play up for days... Then for some reason it started doing it daily. Very strange.

Wish me luck

Threepwood · 2025-07-21T12:43:11+0100

BIOS up to date? the current bios might not like your ram

deviation · 2025-07-21T15:27:30+0100

Threepwood said:
BIOS up to date? the current bios might not like your ram

Unfortunately it is, dude; however, it is a beta version due to when AMD had that scare with:

CVE-2024-36877 security issue, AGESA ComboAm4v2PI 1.2.0.Cc update - Added security issue of SMM Lock Bypass uCode fix aka“Sinkclose”.

They haven't updated it since the end of September

I was quick to update it at the time due to this, so it's strange that it would randomly play up now if it was buggy, although not impossible, I guess.

I did think that could potentially be an issue though, but there's nothing I can do on that front, until they bother, if at all, to update it

Currently it's passed 9 runs of Memtest86+ with XMP re-enabled

so it might have just decided to play nice now if it is the BIOS being mischievous

, or my 4th reshuffling of the cables, and ram modules, last night has fixed something :cry:

I shall leave it going till I'm home later and give it the full 9-12 hours, after that I think it's safe to say the RAM is probably fine?

Either way if it's now fixed and behaves, I'm a happy chappy

aaronyuri · 2025-07-21T15:51:23+0100

If you can run Memtest for 12 hours then I'd say the RAM is good.

TIme to let your usage of it be the next stability test, me thinks!

deviation · 2025-07-21T16:28:52+0100

aaronyuri said:
If you can run Memtest for 12 hours then I'd say the RAM is good.

TIme to let your usage of it be the next stability test, me thinks!

Yep, 12 hours has always been, a reasonable amount of time for me

Yeah, that's the part I'm a bit less excited about

as it was some random when it would do it, then started occuring more frequently, there was no sense to it, and you couldn't make it do it on demand?
Hopefully it behaves now