[SOLVED] MB or PSU on Win10 Uncommanded Restarts...?

Tigerhawk30

Distinguished
Dec 16, 2015
221
15
18,765
Greetings all!

I've been having an issue that I've been trying to fix and I believe I've eliminated every other software possibility, so I'm down to hardware at this point.

I've read that Windows uncommanded restarts that are generated by the Event Viewer that show "Source: Kernel Power", "Event ID: 41", "Task Category 63" can be a failing power supply. I have a replacement on order and should be here in a few days.

However, I'm wondering if this might somehow also relate to a failing motherboard. I've never seen anything that says it might, but sometimes PCs gonna PC and I just wanted to either confirm or eliminate this possibility of a bum motherboard also potentially causing this error set.

Sometimes, the computer will restart itself 10 minutes after a POST, sometimes it'll be 10 hours and sometimes not do it at all. One thing I consistently do to draw power is use it for F@H, otherwise gaming here and there. BIOS and chipsets are updated to the latest versions. Win10 is up to date.

Thanks in advance!
 
Last edited:
Solution
Learning something new every day...I'd never heard of that before. Thanks much for that pointer, I will give that a go later tonight. And you're correct, I've never once reset a CMOS before, ever, in any computer I've ever owned. Thus never having heard of or considered it.

I'd mainly asked this question (PSU or MB) since a friend of mine does IT for a living and that was his continual experience with the code. But if doing the CMOS thing works, I'll certainly give you a thousand thanks for the simplicity of it vs doing surgery to replace a PSU.

Thanks for the education! I will try it later on and report back with what it looks like. I must say though, your thoughts make more sense than a bum PSU since it doesn't shut down...

Tigerhawk30

Distinguished
Dec 16, 2015
221
15
18,765
Are you overclocking? CPU or memory...even if using XMP.

Try resetting CMOS and see if it works acceptable when running in pure stock configuration.

No overclocking at all/ever on the CPU or GPU. I don't see the point, personally.

So, CMOS has something to do with uncommanded restarts relating to the Event Viewer where a power failure is concerned...?
 
....
So, CMOS has something to do with uncommanded restarts relating to the Event Viewer where a power failure is concerned...?
If there are illogical BIOS settings resident in CMOS it can have everything to do with it.

Resetting CMOS is a simple thing to do. It ensures the system is operating in a full-stock configuration so that troubleshooting can proceed from a known stable configuration. It's especially important to do it if you've never done it before.

FYI: a Kernel Power Failure event ID 41 will be recorded in the event log mainly when there is an improper shutdown of Windows from any cause. The improper shutdown is itself a symptom of the problem, not the problem. The problem might be a failing PSU but you're nowhere close to leaping to that conclusion based on information in the OP.
 
Last edited:

Tigerhawk30

Distinguished
Dec 16, 2015
221
15
18,765
If there are illogical BIOS settings resident in CMOS it can have everything to do with it.

Resetting CMOS is a simple thing to do. It ensures the system is operating in a full-stock configuration so that troubleshooting can proceed from a known stable configuration. It's especially important to do it if you've never done it before.

FYI: a Kernel Power Failure event ID 41 will be recorded in the event log mainly when there is an improper shutdown of Windows from any cause. The improper shutdown is itself a symptom of the problem, not the problem. The problem might be a failing PSU but you're nowhere close to leaping to that conclusion based on information in the OP.

Learning something new every day...I'd never heard of that before. Thanks much for that pointer, I will give that a go later tonight. And you're correct, I've never once reset a CMOS before, ever, in any computer I've ever owned. Thus never having heard of or considered it.

I'd mainly asked this question (PSU or MB) since a friend of mine does IT for a living and that was his continual experience with the code. But if doing the CMOS thing works, I'll certainly give you a thousand thanks for the simplicity of it vs doing surgery to replace a PSU.

Thanks for the education! I will try it later on and report back with what it looks like. I must say though, your thoughts make more sense than a bum PSU since it doesn't shut down like that all the time in any kind of a regular pattern.
 
Learning something new every day...I'd never heard of that before. Thanks much for that pointer, I will give that a go later tonight. And you're correct, I've never once reset a CMOS before, ever, in any computer I've ever owned. Thus never having heard of or considered it.

I'd mainly asked this question (PSU or MB) since a friend of mine does IT for a living and that was his continual experience with the code. But if doing the CMOS thing works, I'll certainly give you a thousand thanks for the simplicity of it vs doing surgery to replace a PSU.

Thanks for the education! I will try it later on and report back with what it looks like. I must say though, your thoughts make more sense than a bum PSU since it doesn't shut down like that all the time in any kind of a regular pattern.
It seems that CMOS resets have become more important with Ryzen. I can't be positive, but DDR4 memory training seems to be a part of it: the IMC (part of the CPU) learns what works best for the memory it finds and saves that in CMOS so it can initialize faster the next bootup. As well, the fact that AM4 socket is so long-lived probably has a lot to do with it: with so many CPU generations that work on it the BIOS has lot of configurations to keep track of and needs some help. Making it start with a clean slate, to fully initialize for a new system is a simple way to help it.

One other thing: since you have been having improper shutdowns, aka 'crashes', a good thing to do is to open up a command prompt with admin rights and issue "sfc /scannow". That instructs the OS to do a system file check and attempt to repair any corruption it finds. Do it after resetting CMOS then use the system a while.
 
Last edited:
  • Like
Reactions: Tigerhawk30
Solution

Tigerhawk30

Distinguished
Dec 16, 2015
221
15
18,765
It seems that CMOS resets have become more important with Ryzen. I can't be positive, but DDR4 memory training seems to be a part of it: the IMC (part of the CPU) learns what works best for the memory it finds and saves that in CMOS so it can initialize faster the next bootup. As well, the fact that AM4 socket is so long-lived probably has a lot to do with it: with so many CPU generations that work on it the BIOS has lot of configurations to keep track of and needs some help. Making it start with a clean slate, to fully initialize for a new system is a simple way to help it.

One other thing: since you have been having improper shutdowns, aka 'crashes', a good thing to do is to open up a command prompt with admin rights and issue "sfc /scannow". That instructs the OS to do a system file check and attempt to repair any corruption it finds. Do it after resetting CMOS then use the system a while.

CMOS has been reset. Now for the marathon test...

Thanks much for this pointer. Hopefully this fixes it. I'll be back in a few days to report back.
 

Tigerhawk30

Distinguished
Dec 16, 2015
221
15
18,765
It seems that CMOS resets have become more important with Ryzen. I can't be positive, but DDR4 memory training seems to be a part of it: the IMC (part of the CPU) learns what works best for the memory it finds and saves that in CMOS so it can initialize faster the next bootup. As well, the fact that AM4 socket is so long-lived probably has a lot to do with it: with so many CPU generations that work on it the BIOS has lot of configurations to keep track of and needs some help. Making it start with a clean slate, to fully initialize for a new system is a simple way to help it.

One other thing: since you have been having improper shutdowns, aka 'crashes', a good thing to do is to open up a command prompt with admin rights and issue "sfc /scannow". That instructs the OS to do a system file check and attempt to repair any corruption it finds. Do it after resetting CMOS then use the system a while.

No go. Got the restart three hours in.
 
No go. Got the restart three hours in.
Well, now is when you start eliminating possibilities.

Try one memory DIMM at a time and in different slots (if in A2/B2, try A1/B1, for instance).

All the while, keep BIOS settings in the full default following a CMOS reset. It won't be the optimum but if it's not stable in full default there's no sense pushing performance until you find the defective hardware.
 

Tigerhawk30

Distinguished
Dec 16, 2015
221
15
18,765
Well, now is when you start eliminating possibilities.

Try one memory DIMM at a time and in different slots (if in A2/B2, try A1/B1, for instance).

All the while, keep BIOS settings in the full default following a CMOS reset. It won't be the optimum but if it's not stable in full default there's no sense pushing performance until you find the defective hardware.

Going to bed at this moment, but had already run a 24+ hour Memtest with no faults detected. But, will try this in the morning (if I remember...lol)
 

Tigerhawk30

Distinguished
Dec 16, 2015
221
15
18,765
Well, now is when you start eliminating possibilities.

Try one memory DIMM at a time and in different slots (if in A2/B2, try A1/B1, for instance).

All the while, keep BIOS settings in the full default following a CMOS reset. It won't be the optimum but if it's not stable in full default there's no sense pushing performance until you find the defective hardware.

I'd noted/remembered your comment on BIOS default settings this morning, so I'd lowered the RAM back down to 2666 this morning in the BIOS from the 3600 that the RAM is capable of. I'd just read this morning that the maximum is 3200 with a 5950X...which is very different than the 3600 compatibility I've read in a lot of different places prior...and even the guys at Micro Center, when I bought the RAM at that speed, commented about how I must have a Ryzen because 3600 was supposed to be the sweet spot. So...we'll see where this goes and if it holds up.
 
I'd noted/remembered your comment on BIOS default settings this morning, so I'd lowered the RAM back down to 2666 this morning in the BIOS from the 3600 that the RAM is capable of. I'd just read this morning that the maximum is 3200 with a 5950X...which is very different than the 3600 compatibility I've read in a lot of different places prior...and even the guys at Micro Center, when I bought the RAM at that speed, commented about how I must have a Ryzen because 3600 was supposed to be the sweet spot. So...we'll see where this goes and if it holds up.
While 3200 is the rated memory speed for all Zen 2 and 3 CPU's they are generally quite capable of 3600, but not by just setting the clock speed. Not even always by turning on XMP (or DOCP for Asus). A lot depends on the particular memory DIMM's.

Also, using 4 DIMM's is always a bit more difficult if you have that many.
 
Last edited:

Tigerhawk30

Distinguished
Dec 16, 2015
221
15
18,765
So, as you've said prior, it may not be optimal but I ended up unchecking the automatic restart button in the Control Panel option and I've had no further restarts for three days. This is a clean install of Win10 as of two weeks ago and had actually done the sfc/scannow several times prior to that; it had corrected things a couple of months before the clean install, but yielded nothing further after that.

I'll keep an eye on things and do the maintenance (such as the scannow) here and there to make sure things are in working (enough) order. Gave you a Best Answer in any case for the info that I did not have prior to your suggestions.

Thanks much for all of your pointers, and have a great day!
 

Tigerhawk30

Distinguished
Dec 16, 2015
221
15
18,765
While 3200 is the rated memory speed for all Zen 2 and 3 CPU's they are generally quite capable of 3600, but not by just setting the clock speed. Not even always by turning on XMP (or DOCP for Asus). A lot depends on the particular memory DIMM's.

Also, using 4 DIMM's is always a bit more difficult if you have that many.

Final update: It was the PSU...sort of.

I finally went and swapped out the PSU for my machine. Turns out that the 24-pin ATX connector on the PSU side didn't "click" in all the way. I wouldn't have discovered this if I hadn't gone in to do a swap out. My working theory is that the PSU side of the cable would ever so slightly become disconnected due to the lack of the "click". I've since replaced with the set of cables I know work and have clicked in firmly. If anything changes, I'll update again.

In this case...it was one of the little things.

Again, thanks for your help with suggestions!

Have a great day!
 

TRENDING THREADS