"Random" shut offs

[grim9x8]

Hey friends I only talk to when something goes wrong.

What's happening: I have been experiencing random shut offs the past couple of days. Not crashes full fans stop spinning shut off. after powering down it will attempt to boot back up but will fail, it repeats this accouple times before giving up booting but still turning on enough for my led ram and mother board to kick on. 

symptoms: I ran a log on OpenhardwareMonitor hoping to find a reason for the shut downs but nothing anomalous showed up. temps are fine on everything, im not familiar with alot of the read outs but nothing showed a noticeable spike at the time of the shut down. In event viewer I receive a critical event "kernal power" at the time of shut down

when attempting to boot back on its own my mother board has a series of lights cpu, dram, vga, boot which light up in order during boot, but during these failed boots it only makes it to the first one CPU.
Trouble shooting: checked for windows update, ran Ccleaner, and malware bytes, updated drivers for my gpu, checked my cable connections reseated my ram and gpu, replaced my thermal paste on my cpu. 

So im kinda out of ideas any help?

 

Specs

• OS: windows 10
• mother board: ASUS PRIME Z690-A
• CPU:Intel Core i7-12700K
• Ram: 32gb ddr5 corsair dom plat
• GPU: NVIDIA GeForce RTX 3080
• PSU EVGA 1000w

 

[Bombastinator]

That would normally sound like a PSU OCP problem, but You should have enough PSU there. You’ve got a couple serious space heater components there though, and the time thing might be cooling time, so it could be a heat problem.  Any information in the error logs?

[grim9x8]

16 minutes ago, Bombastinator said:

That would normally sound like a PSU OCP problem, but You should have enough PSU there. You’ve got a couple serious space heater components there though, and the time thing might be cooling time, so it could be a heat problem.  Any information in the error logs?

The highest Temp in the cpu log is 39c across cores is that too high?
the psu is kind of old now I've had it since like 2014.

the following is every error/ warning around the kernal power event

• Error setting traits on Provider {8444a4fb-d8d3-4f38-84f8-89960a1ef12f}. Error: 0xC0000001
• The previous system shutdown at 10:21:02 PM on ‎12/‎15/‎2022 was unexpected.
• The speed of processor 16(17,18,19) in group 0 is being limited by system firmware. The processor has been in this reduced performance state for 71 seconds since the last report.

[Bombastinator]

42 minutes ago, grim9x8 said:

The highest Temp in the cpu log is 39c across cores is that too high?
the psu is kind of old now I've had it since like 2014.

the following is every error/ warning around the kernal power event

• Error setting traits on Provider {8444a4fb-d8d3-4f38-84f8-89960a1ef12f}. Error: 0xC0000001
• The previous system shutdown at 10:21:02 PM on ‎12/‎15/‎2022 was unexpected.
• The speed of processor 16(17,18,19) in group 0 is being limited by system firmware. The processor has been in this reduced performance state for 71 seconds since the last report.

No that’s really really low. Like “are you on open loop?” low.  There’s a test... Heh… Prime95 and furmark are old school power viruses. They will push your cpu (for Prime95) or gpu(for furmark) as hard as it is possible to push them and more if they can.  They were used as old school stress tests.  They aren’t that useful as benchmarks, especially anymore, but they’re great for wasting electricity and making heat. You want to see how hot your machine can get those will do it. If you ram say Prime95 and got a shutdown it might say something useful.  Likewise for furmark, mostly because one does cpu but is very light on gpu and the other is the reverse.  They can be useful tools if you’re chasing a stress based bug.

[grim9x8]

14 minutes ago, Bombastinator said:

No that’s really really low. Like “are you on open loop?” low.  There’s a test... Heh… Prime95 and furmark are old school power viruses. They will push your cpu (for Prime95) or gpu(for furmark) as hard as it is possible to push them and more if they can.  They were used as old school stress tests.  They aren’t that useful as benchmarks, especially anymore, but they’re great for wasting electricity and making heat. You want to see how hot your machine can get those will do it. If you ram say Prime95 and got a shutdown it might say something useful.  Likewise for furmark, mostly because one does cpu but is very light on gpu and the other is the reverse.  They can be useful tools if you’re chasing a stress based bug.

the 39 was just playing a card game+ whatever i was doing on chrome, and the shut downs arnt really correlating to stress as far as i can tell. 

i ran acouple tests in aida64 but non prompted a shut down. as for cooling im using a nh-d15 which is nice but nothing crazy

[Cyberat]

Oh boy, only 68cfm for CPU cooling. You have a CPU overheating problem. For starters go to Bios and set CPU fan to 100% at 60C.

Though not so quick to overlook a dying PSU.

[Bombastinator]

2 hours ago, Cyberat said:

Oh boy, only 68cfm for CPU cooling. You have a CPU overheating problem. For starters go to Bios and set CPU fan to 100% at 60C.

Though not so quick to overlook a dying PSU.

There’s also a possible mount problem in that case.  When was the cooler last off the cpu?

[grim9x8]

9 hours ago, Cyberat said:

Oh boy, only 68cfm for CPU cooling. You have a CPU overheating problem. For starters go to Bios and set CPU fan to 100% at 60C.

Though not so quick to overlook a dying PSU.

I don't follow where are you getting 68cfm from?

[Bombastinator]

3 hours ago, grim9x8 said:

I don't follow where are you getting 68cfm from?

Probably from looking up your cooler. A heavily overclocked 12700k can be a very hot boy.  It’s why I asked about temps.  I suppose it’s possible that the cooler is good but you’re overwhelming it.  Do you have any kind of OC set?  If so it just might be an unstable overclock.  12700 with no overclock is a sub 100w chip though.  The 12700 is real world 65w. Adding a k traditionally increases heat levels by 20%

 

I’ve got a 12700 on a b660 that I thought I could overclock via the external clock method.  I had bad info and bought the wrong board.  I found I had zero use for an OC though. I don’t notice the difference between fss on or off, so I runntiny Tina at 1080p & blow it up to 4k a 100fps lock.  My .1% lows are still over 99fps.  Smooth as butter.  CPUs don’t care about Rez though, so I’ve got no reason to overclock because it doesn’t change my fps.

Edited December 16, 2022 by Bombastinator

[grim9x8]

35 minutes ago, Bombastinator said:

Probably from looking up your cooler. A heavily overclocked 12700k can be a very hot boy.  It’s why I asked about temps.  I suppose it’s possible that the cooler is good but you’re overwhelming it.  Do you have any kind of OC set?  If so it just might be an unstable overclock.  12700 with no overclock is a sub 100w chip though.  The 12700 is real world 65w. Adding a k traditionally increases heat levels by 20%

 

I’ve got a 12700 on a b660 that I thought I could overclock via the external clock method.  I had bad info and bought the wrong board.  I found I had zero use for an OC though. I don’t notice the difference between fss on or off, so I runntiny Tina at 1080p & blow it up to 4k a 100fps lock.  My .1% lows are still over 99fps.  Smooth as butter.  CPUs don’t care about Rez though, so I’ve got no reason to overclock because it doesn’t change my fps.

I haven't over clocked at all, and there's no evidence its running hot.

[Bombastinator]

1 hour ago, grim9x8 said:

I haven't over clocked at all, and there's no evidence its running hot.

I’d say maybe do a memory check.  With memory it’s normally blue screens not restarts, but it’s free.  couple full runs of memtest86 perhaps.  It’s quite slow and gets slower the more memory it has to check.  It’s automatic though.  Something to set and go to bed.

 

A PSU tester won’t do you a heck of a lot of good.  Even if it’s the PSU acting up, you’d have to catch it while it is doing it and it doesn’t sound like an easily reproducible problem.  The kinds of equipment that can do things like graph over time and do continuous testing is not consumer grade.  How much trouble would digging up a PSU to try swapping out be?

[Mark Kaine]

could be ram, try running just 1 stick at a time... test all slots separately,  no xmp etc 

 

also update chipset and bios

[grim9x8]

19 hours ago, Bombastinator said:

I’d say maybe do a memory check.  With memory it’s normally blue screens not restarts, but it’s free.  couple full runs of memtest86 perhaps.  It’s quite slow and gets slower the more memory it has to check.  It’s automatic though.  Something to set and go to bed.

 

A PSU tester won’t do you a heck of a lot of good.  Even if it’s the PSU acting up, you’d have to catch it while it is doing it and it doesn’t sound like an easily reproducible problem.  The kinds of equipment that can do things like graph over time and do continuous testing is not consumer grade.  How much trouble would digging up a PSU to try swapping out be?

The shut offs became more frequent so I tried test individual ram sticks, and now we can't boot at all. The dram light is remaining on on the mother board. The manual claims of a led remains lit it means an error with the component. 

 

I think permanent shut down was more coincidentally with the slot swaps than caused by it.

 

I think I'm going to buy the same ram again and replace these sticks if the problem persists then I at least can still use the old ram with the new ones if they're matching.

 

Read some posts describing my new symptoms and they thought it was their mother board but it wasn't very conclusive.

 

[Bombastinator]

9 hours ago, grim9x8 said:

The shut offs became more frequent so I tried test individual ram sticks, and now we can't boot at all. The dram light is remaining on on the mother board. The manual claims of a led remains lit it means an error with the component. 

 

I think permanent shut down was more coincidentally with the slot swaps than caused by it.

 

I think I'm going to buy the same ram again and replace these sticks if the problem persists then I at least can still use the old ram with the new ones if they're matching.

 

Read some posts describing my new symptoms and they thought it was their mother board but it wasn't very conclusive.

 

Intel 12th gen is often limited to 3200 MHz memory.  I need to find out if 13th gen will take faster memory or not. My board will do over 3200mhz but apparently my cpu won’t.  a lot of AMD stuff is effectively though not actually limited to 3600mhz memory at the top end.  The fact that Intel stuff is stuck at 3200 though makes that speed so much cheaper that efficiency dictates that speed.  The red light is a POST problem. It could be that you just didn’t put the memory back in again right (manual will know)  It does indicate that there may be a problem with the RAM though.

 

UPDATE:  it apparently isn’t exactly, or at least isn’t necessarily.  Both AMD and Intel are guaranteed to do 3200 MHz. But AMD stuff can effectively do up to 3733 before the infinity fabric limitations kick in.  If you overclock them it can go higher, but if you go too fast the infinity cache switches from 1:1 to 1:2 and things actually get slower for a bit.   If your processor will do it.  Intel has no infinity fabric issue so they can theoretically go higher, (though things get radically more expensive and again, it’s not guaranteed) so that $250 4800mhz memory may not work. Or it may. It’s luck, apparently, unless you have an h series board.  Then iirc it’s 2666, or something. 

Edited December 18, 2022 by Bombastinator

[IkeaGnome]

9 minutes ago, grim9x8 said:

The shut offs became more frequent so I tried test individual ram sticks, and now we can't boot at all. The dram light is remaining on on the mother board. The manual claims of a led remains lit it means an error with the component. 

 

I think permanent shut down was more coincidentally with the slot swaps than caused by it.

 

I think I'm going to buy the same ram again and replace these sticks if the problem persists then I at least can still use the old ram with the new ones if they're matching.

 

Read some posts describing my new symptoms and they thought it was their mother board but it wasn't very conclusive.

 

Looks like you're running DDR5? Have you updated BIOS to the most recent, and updated chipset drivers?

Since the original bios release there have been 5 releases improving ram stability and compatibility. Just note that for the most recent BIOS, you'll have to update Intel ME with Asus' utility. 

[grim9x8]

1 hour ago, IkeaGnome said:

Looks like you're running DDR5? Have you updated BIOS to the most recent, and updated chipset drivers?

Since the original bios release there have been 5 releases improving ram stability and compatibility. Just note that for the most recent BIOS, you'll have to update Intel ME with Asus' utility. 

Can this be done when not booting?

[IkeaGnome]

26 minutes ago, grim9x8 said:

Can this be done when not booting?

No, but it was suggested long enough ago and you didn't say anything to it so I was wondering if you did try that and it made it worse, or if you didn't try it. 

[grim9x8]

1 hour ago, IkeaGnome said:

No, but it was suggested long enough ago and you didn't say anything to it so I was wondering if you did try that and it made it worse, or if you didn't try it. 

worth doing but probably not the issue at the current moment.

[Cyberat]

Buy Hype, Cook CPU, ooops not posting... Everyone applaud the Noctua hype.

Don't listen to me, what do I know ? I'm not a trendy fashion guy.

[Bombastinator]

39 minutes ago, Cyberat said:

Buy Hype, Cook CPU, ooops not posting... Everyone applaud the Noctua hype.

Don't listen to me, what do I know ? I'm not a trendy fashion guy.

Noctua has good fans.  I they’re not cheap fans though.  They’re also supposed to have a really well done attachment method.  They were also the best at quiet cooling for a lot of stuff prior to 2020.  That was for pre-1700 CPUs though.  You also paid for that in money.  I don’t know if it’s still a thing or not.  The impression I get is it’s more or less open season.  I do know there are cheaper coolers out. I paid like $35 for a big air cooler that apparently doesn’t need fans to keep a 12700 well below 70°c under not terribly heavy load.  I’ve been seeing low to mid 60s actually.  YMMV.  I Haven’t stress tested though.  I get the impression it would still be fine though the temps would be higher. People were liking the scythe fuma a lot for a while but the price on that one started climbing a lot so I don’t know if it’s still a good deal.  The air cooler I bought isn’t even offered anymore.  Weird thing because at $35 it was crazy good.  A less popular cooler may not compete well in its size class, but when the price starts dropping it can turn into a really good buy.

Edited December 18, 2022 by Bombastinator

[Cyberat]

68-82 CFM Is NOT a good thing. Wasn't in the past either as my old AMD FX cpu is 150W TDP.

[Bombastinator]

16 minutes ago, Cyberat said:

68-82 CFM Is NOT a good thing. Wasn't in the past either as my old AMD FX cpu is 150W TDP.

My understanding is CFM isn’t super reliable as a measure for cooler ability.  Fan ability yes.  Lower CFM fans will be less noisy generally, but that’s about as far as it goes.  Mine said it was 68cfm, though now it’s zero cfm because I took the fans off.  I’ve got a 65w chip though.

Edited December 18, 2022 by Bombastinator

[grim9x8]

12 minutes ago, Cyberat said:

68-82 CFM Is NOT a good thing. Wasn't in the past either as my old AMD FX cpu is 150W TDP.

cool man go write a product review im not overheating

[Cyberat]

No, you're not overheating anymore, you overcooked your CPU already.

[Bombastinator]

10 minutes ago, Cyberat said:

No, you're not overheating anymore, you overcooked your CPU already.

Might have happened. Might not.  My understanding is these days CPUs throttle before they can overcook, unless you mean something different. The higher the temperature albedo is the more efficient a cooler is. It seems to be about how well a given cooler works at a given fan speed and a given amount of heat it needs to clear.  Overkill seems to reliably work though.  That’s what I wound up with. I don’t know of a way to get more than 80some cfm without water. 

Edited December 18, 2022 by Bombastinator