• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Question Kernel Event 41 PSU Issue?

sze5003

Lifer
Aug 18, 2012
13,501
355
126
August 13th I installed a 3080Ti upgrading from a 1080Ti. Been playing games again on my PC on max settings and enjoying it so far. Until yesterday. I've been on vacation the past few weeks and yesterday I get back and after turning on the PC and playing about 15 minutes of Car Mechanic Simulator 2021 (don't judge me, its kind of therapeutic), the PC goes dark for a few seconds and then reboots itself.

I see Kernel Power Event 41 in Windows event viewer. EVGA Precision X1 received an update and a firmware update for my GPU yesterday so I figured maybe its a one time thing. I continue playing Psychonauts 2 and CMS 2021 throughout the day without issues.

This morning I decided to run the OCCT bench application and Power Test. Instantly, the PC does the same thing, goes black and then turns itself back on again. Same Kernel Power Event 41 in the Event viewer. Yesterday I ran the same test for a couple of minutes and it did not turn off and reboot. Not sure why this morning it did reboot.

Board is Asus Z370E

Cpu is i9-9900k stock

GPU is EVGA 3080Ti FTW3 Ultra ( using 3 separate PCIE cables for power)

Memory is Corsair DDR4 4 X8GB @ 2666mhz

PSU is a Seasonic X850 modular 80 Plus Gold I purchased in 2017 on Amazon new.

Right now I am running OCCT CPU stress test for 30 minutes and so far so good. Next I will run the GPU only test. I am thinking its the PSU at this point at high loads or spikes.

I contacted Seasonic and they mentioned to unplug and re-plug all the cables which I already had to do twice since my previous card was 2 8 pin and this one is 3 8 pins so I had to disconnect everything and reroute most my cables because I like being neat.

They also mention to try under volt GPU which I've never done so I can look into this. They did mention if nothing helps to fill out the RMA form.

Is there anyway for me to trigger the issue repeatedly as I think at this point it will be random when load occurs.
 

mikeymikec

Lifer
May 19, 2011
14,741
4,922
136
kernel event 41 is the generic "the system wasn't shut down cleanly" event log entry, it's very little to go by on its own. It just means that Windows did not shut down normally, you'd get it if the system spontaneously rebooted, or Windows blue-screened, or the power went off unexpectedly. In PC troubleshooting, it just means that something bad happened and just about anything could have caused it.

The first thing I would do is to find out if Windows is reporting a crash. Nirsoft's BlueScreenView is handy for looking back at past Windows crash logs: https://www.nirsoft.net/utils/blue_screen_view.html

On the crash front, I would disable Windows's very unhelpful feature which is enabled by default whereby instead of actually showing you the crash message, it reboots as quickly as it can so hopefully you won't notice that Windows did a whoopsy.

As you're happy to delve into the Windows event logs, do you have any other errors/warnings that are potentially of interest? Unfortunately WIndows 10 complains in the logs like a crochety old man, even a clean install of Win10 creates at least some warnings in my experience.

Does the computer ever have problems in normal usage (browsing, for example)?

I don't have Win10 to hand at the moment, so I've googled the instructions ( I think they mislabelled the article):
Just follow the first four steps. In future, if Windows crashes and presents a BSOD, you should get to see it. When you're done reading the message, use your computer's reset button or hold the power button down for more than 5 seconds to forcibly switch it off.

I know nothing about this OCCT bench program so I don't know what to suggest there, but after I've tried evidence gathering I'd go back to stress testing but limiting the focus of the testing to specific components to hopefully eliminate one at a time. I can make some suggestions for what I'd do in this respect, though others might say "just do x y and z in OCCT".

PS: This thread probably should be moved to a troubleshooting sub-forum for maximum forum attention. It could still be power related, who knows yet.

PPS: I don't like the sound of this graphics firmware update, the timing mostly. I would suggest googling to see if others have had similar problems with this update. Does EVGA have a user forum? If so try there.
 
Last edited:

sze5003

Lifer
Aug 18, 2012
13,501
355
126
kernel event 41 is the generic "the system wasn't shut down cleanly" event log entry, it's very little to go by on its own. It just means that Windows did not shut down normally, you'd get it if the system spontaneously rebooted, or Windows blue-screened, or the power went off unexpectedly. In PC troubleshooting, it just means that something bad happened and just about anything could have caused it.

The first thing I would do is to find out if Windows is reporting a crash. Nirsoft's BlueScreenView is handy for looking back at past Windows crash logs: https://www.nirsoft.net/utils/blue_screen_view.html

On the crash front, I would disable Windows's very unhelpful feature which is enabled by default whereby instead of actually showing you the crash message, it reboots as quickly as it can so hopefully you won't notice that Windows did a whoopsy.

As you're happy to delve into the Windows event logs, do you have any other errors/warnings that are potentially of interest? Unfortunately WIndows 10 complains in the logs like a crochety old man, even a clean install of Win10 creates at least some warnings in my experience.

Does the computer ever have problems in normal usage (browsing, for example)?

I don't have Win10 to hand at the moment, so I've googled the instructions:
Just follow the first four steps. In future, if Windows crashes and presents a BSOD, you should get to see it. When you're done reading the message, use your computer's reset button or hold the power button down for more than 5 seconds to forcibly switch it off.

I know nothing about this OCCT bench program so I don't know what to suggest there, but after I've tried evidence gathering I'd go back to stress testing but limiting the focus of the testing to specific components to hopefully eliminate one at a time. I can make some suggestions for what I'd do in this respect, though others might say "just do x y and z in OCCT".

PS: This thread probably should be moved to a troubleshooting sub-forum for maximum forum attention. It could still be power related, who knows yet.

PPS: I don't like the sound of this graphics firmware update, the timing mostly. I would suggest googling to see if others have had similar problems with this update. Does EVGA have a user forum? If so try there.
Thanks for the link on how to disable the auto reboot. I've been looking for that feature. I'm used to seeing blue screens anytime there are errors but these two times there were none.

OCCT is just a stress test program that lets you bench and test different components. Similar to Prime 95 but you can choose to stress individual components or for example like the Power Test where I got the shut off and reboot. It stresses both the CPU and GPU at max power from the PSU. I've heard that this test will often tell you if your PSU is going bad.

I finished a CPU stress test and GPU stress test without any shutdowns or reboots. I'll continue poking around, this never happened on idle or normal computer usage. Happened for the first time while gaming since I got the new card then this morning when I ran the combined power stress test.
 

sze5003

Lifer
Aug 18, 2012
13,501
355
126
Would anyone think this is weird that GPU-Z is reporting the third PCIE 8 pin wattage at around 80 to 90w vs the PCIE #1 and #2 are showing 120w or more? The reboots seem to be very random and often always occur when I run the OCCT tool Power test after a clean reboot. If I run the same stress test after the computer is back up from the crash, it runs fine. Makes me think that randomly when there are power spikes, the PSU could be tripped and that could be causing the reboot. For reference this is the tool I am using: https://www.ocbase.com/
 

Attachments

aigomorla

Cases and Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
19,196
1,589
126
I just ran into a simular issue with my 3090.
It seems like it can massively pull current from the PCI-E rails which triggers over volt protection on the PSU.

I just submitted an RMA to seasonic.
But i highly doubt 1200W PSU is not enough for a 3960X + RTX 3090.

But my issue is repeatable, and if i lower the power settings on my EVGA precision X1 i do not get the random restarts when playing a game, so i am confident i am somehow triggering my over volt protection on my seasonic prime.
 

sze5003

Lifer
Aug 18, 2012
13,501
355
126
I just ran into a simular issue with my 3090.
It seems like it can massively pull current from the PCI-E rails which triggers over volt protection on the PSU.

I just submitted an RMA to seasonic.
But i highly doubt 1200W PSU is not enough for a 3960X + RTX 3090.

But my issue is repeatable, and if i lower the power settings on my EVGA precision X1 i do not get the random restarts when playing a game, so i am confident i am somehow triggering my over volt protection on my seasonic prime.
I don't think it's an issue of the PSU not being enough in terms of wattage.

It's that the older seasonic models are having their ocp tripped. I heard their new line or their refurbished models have been updated against this.

I have been posting on the evga forums and if you search there, there are so many posts about the same issue we are seeing here.

For me I can replicate it randomly when I play too much of car mechanic simulator or when I randomly run the OCCT Power Test after the pc has been under some load for some time.

I ordered an EVGA G6 1000w power supply. It's their 2021 model so it should have better tolerances for the ocp protection.

The Corsair RMX models are also ones that are 2021 models and users have not reported any issues with them.

Seasonic offered me an RMA but they would probably said they would give me a focus gx 850 or focus prime 850 or the same model I have but refurbished.

Not sure if any of these models have been updated for the power spikes of these cards. I'll probably just sell whatever I get back from the RMA.
 
Last edited:

aigomorla

Cases and Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
19,196
1,589
126
well i'll let you know what they respond to me.
My current PSU is a seasonic platinum SS-1200W

If they do not offer me another platinum psu, i'll probably get saucy.

I was also looking at eVGA supernova platinum 1300W P+

I may end up just getting that and waiting on the RMA, as i have too much expensive stuff on the line to wait it out comfortably.
 

sze5003

Lifer
Aug 18, 2012
13,501
355
126
Here is one of the many EVGA threads about PSU's and 30 Series cards:


I went back and forth with Seasonic support several times asking what I would get back if I did an Advanced RMA. You will get two options, regular RMA and Advanced.

What you get back is dependent on stock but I don't have faith that the models they may send out are the fixed ones or newer models rebuilt to withstand the high power draws or transient loads.

I even asked if they could provide me a 1000w model that they know is working with the 30 series cards in terms of load spikes, because I'm sure they are aware of the issue, I would pay the difference in cost but they said they would not do that for a warranty claim.
 

aigomorla

Cases and Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
19,196
1,589
126
FML.... whatever is on stock is not sitting too well with me.
If i have a platinum, i better get another platinum.

Other then that spike, i have no issues with this PSU.
I may just end up keeping it, and grabbing the EVGA.

Only issue is that no authorized vendor has it.
Its all third party which pisses me off, as i want to make sure i get the full warrenty for it.

EDIT:
So newegg magically got one in stock, and i ordered it.
EVGA Supernova P+ 1300W. Its coming tomorrow, but i wont have a way to test it until new world launches, and i probably won't test it that way until after i get a decent head start in the game.

But i can always move the seasonic to another system, possibly my mining rig, as it had no problems with anything else but my 3090.

I'll let you know how that PSU handles, but im only hearing good things about it, as its made by Super Flower, and almost on the same platform as the leadex series.
 
Last edited:

sze5003

Lifer
Aug 18, 2012
13,501
355
126
Stock was my issue too when I made the post on the EVGA forums.

The supernova G6 1000w I ordered should be here on the 15th but I may not get to installing it until Sunday.

It was normally $220 but EVGA had it on sale for $169 and I also got another 10% off with an associate's code.

Oh and be careful with new world, there's a huge post on EVGA as well about that game bricking 30 series cards because it seems most models heat up too much and have a soldering issue.

I thought it was singled out to 3090's but someone with a 3080ti also had the card go dead after trying to play that game.
 
Last edited:

aigomorla

Cases and Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
19,196
1,589
126
mines water cooled and i put afermarket heatsinks on the backplate to keep that under check even.
The rear hotshot doesn't go past 64C while playing that game. unless my room is a sauna, and i do not turn on the AC.
 

ASK THE COMMUNITY

TRENDING THREADS