Kernel-Power 41 Task 63 Shutdowns Since Upgrading to a 3080.

[domchu]

Hi all, as the title states, I've been experiencing shutdowns only when my system is idling/under light load - it has never happened during heavy loads. Every time it crashed, I would launch event viewer and it would always show 'Kernel-Power 41 Task 63', and at this point I've lost count. This has been happening ever since I upgraded from an RTX 2060 to a 3080 and I've never had this kind of shutdown in the 20 months I've had my 2060 before.

 

I've seen many forums where people had experienced this kind of problem, and many believe the PSU is at fault, which is possible but in my case, I doubt it. I've been using my new PSU (from be quiet! DPP11 to Seasonic TX1000) for three months before I upgraded from a 2060 to a 3080 which is when the crashes started occurring.

 

I've reset Windows 10 twice and it still happens, I don't know whether its a software or hardware problem since it continued after two full system resets.

 

I don't have much else to say but I would appreciate a solution to this problem, and if any of you are experiencing the same issue, feel free to voice your situation in the comments.

[Pwnstix]

Sorry to revive this old thread, but I just recently found it. 

 

I've been experiencing the same issues with my RTX 3080 (EVGA FTW3 Ultra) and TX-1000 PSU. I bought the PSU and installed it at the same time as my RTX 3080, upgrading from a Seasonic Prime Titanium 750W PSU that was only ~3 years old. I got it the 1000W PSU just in case I'd have issues with the 750W PSU. Everything seemed fine, at first, but whenever I played Cyberpunk 2077, sometimes, but not always, my PC would reboot spontaneously whenever I paused the game.

 

At first, I thought it was the just the game, but I decided to run OCCT's 3D stress test, and that would often trigger the same rebooting issue. So, started a long process of trying to figure out what the issue was. I tried the usual: resetting any overclocks to default settings in both my CPU and GPU, turned off XMP, turned off any overlays or onscreen displays. Ran a full memtest86 and passed. I made sure my drivers were updated, and I also used DDU to uninstall my graphics driver and tried clean installing the latest ones, and also tried rolling back to older ones. Nothing seemed to work, so I ended up swapping my TX-1000 out and putting my older 750W PSU back in. The same issues occurred. So I put the 1000W back in. I also use an uninterruptible power supply (UPS), and I'd just got a new 1000W UPS, so it is also fairly new. But I tried plugging directly into the wall, bypassing the UPS. I thought that had fixed it, but eventually it started happening again. 

 

I ended up RMAing my RTX 3080 and got a replacement, what looks to be brand new and not refurbished. It started happening again. The rebooting also started to occur in other games, besides the high powered Cyberpunk, such as Heroes of the Storm, Battlefront 2, and even Europa Universalis IV, which is not a very demanding game. 

 

After going through the EVGA forums, I came to a sort of conclusion that the EVGA cards, and perhaps all NVidia 30 series cards, have an issue where they draw way too much power for very quick periods of time, and this can trip up some PSUs...something about transient power spikes, transient response, etc. But I had read, though never confirmed it for myself, that the TX-1000 was supposed to have had its transient response fixed by Seasonic, and I'd seen that other people were using this PSU, so I thought that it couldn't possibly be the power supply. 

 

The rebooting issues persisted and seemed to get worse, and once, I found that my PC had shut off some time while it was idle. I considered that it might be a motherboard or CPU issue...so I ended up building a new PC. I switched from an Intel i7 8700K/Z370 motherboard to what I have now (AMD Ryzen 7 5800X/X570 motherboard) and also got new RAM. I also clean installed Windows. The RTX 3080 and TX-1000, of course, I kept. I was really hoping that would solve the problem.

 

I've had this new build now for about 5 days. I was running OCCT's 3D stress test every day and passing---except for yesterday, when my PC shut down. So, now I'm convinced that it was probably the PSU all along. I still think the video card is part of the problem, and this is still a good PSU, but I think there's something in it that's not handling transient power spikes as well as it could, or perhaps as well as another unit might be able to. Unfortunately, there is currently zero stock of power supplies out there that aren't being scalped, so I cannot buy another one to try out. And if I RMA this TX-1000, I'm sure Seasonic will probably send me the same model, and it would really suck if it has the same issues. 

 

Again, sorry for resurrecting this post, and for my long post...but the OP asked if anyone else was having a similar issue with this particular PSU (TX-1000), and here I am. I'd also like to hear from the OP -- have you managed to figure anything out on your end? I'm guessing if you're in need of another PSU, then you're in the same boat as me, as PSU stock seems to have been dry as far back as December 2020...

 

Anyway, I'll update if I do manage to get ahold of another new PSU. I'm looking into the Corsair AX1600i...which I know is completely overkill for my usage, but all the reviews I've seen have shown that it has great transient response, in addition to being very efficient. The AX1200i and HX1200i don't get as good reviews for transient response. But yeah none of that matters since they're apparently all out of stock in the US. 

[Lyri]

Had a Kernel Power Error 41 myself after switching motherboards. In my case the problem was caused by RAM incompatibility with the current motherboard and the problem only persisted when the XMP was turned on. Switched RAM, problems disappeared. See if that helps. @domchu @Pwnstix

[Pwnstix]

@Lyri Yeah I've switched motherboards and RAM (and CPU) at this point, and I've got the same kernel-power error 41 task 63 events as @domchu

 

I'm pretty sure my issue with with my power supply, so I'm looking into getting a new one. I may downgrade to an 850W PSU, but a known good PSU for the 30 series video cards with excellent transient response, such as an EVGA 850W P2. It's built on the Superflower Leadex platform, which is highly rated. 

[Lyri]

9 hours ago, Pwnstix said:

@Lyri Yeah I've switched motherboards and RAM (and CPU) at this point, and I've got the same kernel-power error 41 task 63 events as @domchu

 

I'm pretty sure my issue with with my power supply, so I'm looking into getting a new one. I may downgrade to an 850W PSU, but a known good PSU for the 30 series video cards with excellent transient response, such as an EVGA 850W P2. It's built on the Superflower Leadex platform, which is highly rated. 

Could be PSU, yeah. I did have people telling me that. It is a really nasty BSOD to be having, I ended up pretty much just swapping every single component in my setup until I stumbled upon RAM. Funny enough, I even had to install a different version of Windows for the problems to finally be over, so don't rule out software sided issues entirely.

[Pwnstix]

10 hours ago, Lyri said:

Could be PSU, yeah. I did have people telling me that. It is a really nasty BSOD to be having, I ended up pretty much just swapping every single component in my setup until I stumbled upon RAM. Funny enough, I even had to install a different version of Windows for the problems to finally be over, so don't rule out software sided issues entirely.

Yeah I've swapped out pretty much everything and clean installed Windows, especially since I was switching from Intel to AMD. I know it could technically have worked with my older Windows installation installed on my OS drive, but I didn't want to risk all the weird driver errors/incompatibilities, plus it was probably due for a clean installation. 

 

As for BSODs, I never got any of those. And I had Windows' automatic restart disabled, just in case there were any actual BSODs, but all I got were hard reboots/shutoffs. Actually, just last night, I found my PC had shut off, just like it did in the last week that I was using my old components. The core components that are still in there: the video card and PSU. 

 

I just ordered an EVGA Supernova P2 850. I hope I don't regret downgrading my wattage by 150W, but the 1000W P2s weren't in stock, and I feel like I need to do something. Plus, the P2 850 got really good reviews on multiple sites, particularly in transient tests, which is something I'm paying more attention to, now. 

 

I just wonder if my TX-1000 has some kind of fault, or if it's similar to other high-efficiency PSUs, in that the secondary side (which covers transient response) is sacrificed somewhat in favor of high efficiency. I mean, my older Seasonic Prime Titanium 750W had been doing the same thing with this video card (well technically my first RTX 3080, before I RMA'd it). But that is another high efficiency PSU, that particular model apparently had some issues with transient response, which were later corrected. I thought it was the same with my TX-1000...

 

In any case, I'll find out in a few days if another new PSU fixes this issue. 

[Kasper-DK]

I just made new tread yesterday with almost the same story

 

 

I also came from intel i7 8700K til AMD Ryzen 5800X and also swapped almost all components. Right now I am waiting for RMA status for my Asus RTX 3070 Strix OC. 

 

I just got a comment with suggestion from @ShrimpBrime and I will look into this asap. 

 

Did you receive your new 850W PSU and did it work? 

 

[Pwnstix]

On 5/26/2021 at 9:05 AM, Kasper-DK said:

I just made new tread yesterday with almost the same story

 

 

I also came from intel i7 8700K til AMD Ryzen 5800X and also swapped almost all components. Right now I am waiting for RMA status for my Asus RTX 3070 Strix OC. 

 

I just got a comment with suggestion from @ShrimpBrime and I will look into this asap. 

 

Did you receive your new 850W PSU and did it work? 

 

Whoops, shame on me for not updating this earlier, but I wanted to take a while to make sure everything was working.

 

Yes, it seems like the EVGA Supernova P2 850W I installed did, in fact, fix this issue for me. I haven't actually done a stress test with OCCT yet, because...I'm scared. But I've been playing Cyberpunk 2077 quite a bit, and I haven't had any reboots or shutdowns yet. No idle shutdowns, either. 

 

Just for full disclosure: a few days ago, I did have one weird instance of my PC seemingly having locked up while idle, but hadn't rebooted or shut down. My motherboard (Asus) Q-Code LED display read 'F8,' and I had to hit my reboot button. I looked up that Q-Code and found that it means 'Recovery PPI Not Available,' which wasn't very helpful, beyond it maybe being related to the EDI not initializing, or something...but Googling that, it seems to be related to Ryzen Master, which I had used briefly a day earlier. I didn't make any changes, and I'd closed it afterwards, but I don't know if there are any processes connected to it that might have still been running. Prior to that, I haven't really used Ryzen Master that much, and I've probably manually rebooted after using it anyway. So, I'm not sure if that error was related to Ryzen Master, or if I do still have some kind of other issue.

 

That hasn't happened since, and I'm still not getting any reboots or shutdowns under heavy load. But, like I said, I haven't done any stress testing with OCCT yet. I know I should, though...

 

I sent my TX-1000 off to Seasonic for an advanced RMA. I got a new unit in the mail this past week and I'm waiting to be reimbursed for my collateral (because of the advanced RMA). I'm not sure what I'll do with this replacement TX-1000.

 

Anyway, I'm ~99% sure that there was some issue with the previous TX-1000, maybe related to the secondary side and transient spikes. But I'm not sure I'll ever know for sure, unless this starts happening again. All I know is that I've been relatively problem-free for well over a month.

 

Edit: oh I forgot to mention, I did check my event viewer with I had the weird F8 code on my motherboard, and saw a kernel power event, bit it actually had a bugcheck code and parameters listed, unlike when I had the sudden reboots with no BSOD. I ran it through WinDgb and saw references to video TDR errors, which is a driver-related issue. That sucks, but I think it generally means it's not entirely hardware-related.

[Kasper-DK]

Thank you for your update. Much appreciated.

It sounds like the PSU was be the solution for you. Fingers crossed

 

I have tried to replace PSU from a Corsair RM850 to a Asus Strix 850 (OEM Seasonic), but I did not work out for me. I am still experiencing this error. 

Resently I tried to update BIOS to the lastest version (the BETA version also available for your MB) and it did not work either.

There is some USB related issues to these AMD based mainboards. 

 

Next step for me is replacing all 4 Memory modules to another set/model (2 pcs.) for testing.  

 

Still waiting for my GPU that I have returned for RMA. I am excited to hear if they have detected any error on the GPU. 

 

If replacing RAM not is working as well - I dont know what to do next. 

 

Thank you again for updating. 

 

[Pwnstix]

No problem. Sorry your issues have persisted. I took at a look at your other thread here and it looks like you've ended up replacing even more components than I did. Do you use an uninterruptable power supply, by any chance? Other than that, the only thing I could think of, off hand, would be the wiring in your residence...but I think that being the problem, in cases like this, is usually pretty rare. But sometimes, if there's something power hungry on the same circuit as your PC, it can cause issues. You've probably already ruled that out, though. But I'm just trying to think of any commonality, since you've changed most of the components inside your PC, including all of the core components.

 

It's definitely strange and frustrating...even for me, because I still fear that my issues will pop up again, even if I'm 99.99% certain I've fixed it. 

 

Hopefully somebody more knowledgeable than me can chime in here, or on your own thread if you can bump it.