PC Shuts down completely while gaming and restarts instantly

[TheGodofCruelty]

Hello, I have an issue that started happening more and more over the past few weeks, I am suspecting the PSU but I am not entirely sure, I did some testing and it over time started to happen in games where before it didn't happen, a few months back it happened when I ran a heavily modded sims 3 and started warframe at the same time, I wrote it off as a sims problem, now these last 3 days the PC shut down within 10 mins from starting Baldurs Gate 3 which I could play for literal days without any shutdowns a few weeks ago, also happened in elden ring when a lot of stuff happened at the same time.
Hardware is:

MSI Z390 PRO-A
i9 9900KF non-OC
2x 16G RAM (where I am deeply sorry that I can't for the love of me remember what brand I got since the stickers fell off and I can't remember where I ordered them)
ROG Strix GeForce RTX 2080Ti 11G (which had it's fans replaced, might be another culprit? bad chinese quality parts maybe?)
Corsair RM850x PSU which is a little over 3 years old (it did run 24/7 tho which is why I kinda suspect this being the culprit)
No Mechanical parts other than the Chassis and GPU fans, CPU is Watercooled with an AIO cooler
1x NVMe and 2x Regular SATA SSD

I attached a HWInfo log since I have no clue what to look for exactly, however I do sincerely hope something indicates it's the PSU because if it's the CPU or Board then I am sincerely screwed no way I can afford a new system and getting an LGA1151 board around here is almost impossible given the age...

Edit: One thing I forgot, no errors, Bluescreens or anything in the system logs, only thing that shows up is a critical KernelPower event.
Why do I also suspect the GPU fans as possible culprit? because one of them according to the logs spins at sometimes up to 3 million RPM apparently...

 

[Dalai]

Although it sounds like the PSU being the culprit, I have my doubts about it. A good quality 850W PSU is plenty for such a system, especially considering that the CPU won't be fully loaded while gaming. I suggest narrow it down more to at least try to find out which component is faulty. To do that, stress both CPU and GPU separately with programs like Prime95, OCCT and FurMark - the former two being CPU stress tests and the latter a GPU stress test. Furthermore I suggest to check every cable connection in the system, especially the GPU 8-pins and EPS 12V (CPU) power connectors.

Regards
Dalai

 

[TheGodofCruelty]

I don't get it at all, I ran Prime for half an hour no shutdown,
Furmark right after that with no break in-between, PC immediately turned off the moment It tried to start, window opened, no graphics rendered yet, immediate shutdown.
now I ran it again with HWInfo in the back to possibly get a temp log, this time without Prime95 before it
and now it doesn't shut down even after letting it go through 10 times already...
Seems to only shut down when I either run them shortly after another or if I run both at the same time which also just now Resulted in an instant shutdown after just a few seconds.
Connections are all Stuck as tight as they can be to the connector, checked if something was inside the pins too that didn't belong or if something was bent too but didn't seem like it, reseated the cables too But it still turned off when running both or one after the other...
Kind of only confirmed what I already knew as I was able to AI generate images with my GPU just fine without shutdowns, never tested the CPU tho but seems just fine too.
guess I'll try to shoot into the blue and get a replacement PSU and deshroud my GPU for good measure since the fans are making weird grinding noises anyways.
May as well go for a pay later offer and if that isn't what causes it I'll have the option to return it without having to lay down cash...

 

[Dalai]

Yes, it could be the PSU if the issue occures when both the CPU and GPU are fully loaded. It's just that PSUs from a reputable brand shouldn't make problems at the age of 3 years - though I'm aware that it can happen anyway.

If you happen to have access to a computer shop around the corner or you have a friend who can lend you one, it's probably a good idea to try that before actually buying a new one.

Let's hope it's a culprit that can be solved easily; I hate when issues are not reproducible and I had my fair share of those (hardware and software) ...

Regards
Dalai

 

[TheGodofCruelty]

Eh, possibly just lack of power, 850w after 3½ years, paired with 2 rather power hungry components, a ton Periphery due to VR setup, AiO cooler with quite a few fans (five of them, 3x 140 2x 240)
I can kinda see how after 24/7 usage for that long it might have aged to the point of just not supplying enough power anymore,
that or it might really be one of the cheap chinese replacement fans for the GPU causing issues and sucking too much power,
which is why I'll go ahead and deshroud the GPU, so even if it is the cause with 2 noctua's and new paste that thing should 100% stay cool enough for sure.

But yeah my worst one was crashing and slowdowns / stuttering caused by a plugged in USB Controller dongle if it stayed plugged in during a reboot, pull out, replug, PC ran fine, restart with the dongle, the stuttering began again... weirdest PC issue I had so far...

 

[TheGodofCruelty]

So as it turns out my issue and Paranoia were triggered by two entirely unrelated issues that just happened to show up around the same time frame by pure Coincidence
One were the USB issues that I happened to have which I thought were due to a lack of current or delivered or the board itself.
Everytime I put the system under load, it would disconnect all USB devices for a second and then restart them.
More load and it would just crash the entire system, turns out it was not a sign of a faulty MB but a driver issue with Synapse that just happened to start occuring at around the same time as the PSU issues.
The other thing: I did confirm it was in fact the PSU, swapped that out and no more hard crashes even under loads where everything would grind to a halt, still had the usb thing tho, after performing a complete driver wipe, factory resetting the hardware etc. That issue is seperately solved too so yeah...
TL;DR: PSU was broken, coincidental driver issues made me paranoid about it being an issue with the board
So problem solved.

 

[Dalai]

Thanks for reporting back, and I'm glad that you could solve the issue. Honestly, I still find it weird that such a quite young PSU degraded that quickly... Even if I assume 300W for the GPU and 250W for the CPU (both are on the higher side), add the rest of the components, I consider the total wattage way below 700W. Judging from your tests, the PSU couldn't even deliver 50% of its total wattage at its age.

I suggest to check if the PSU is still under warranty. If it's this model it has a warranty of 10 years, and you could (and should) RMA it. And if it is this model, then that adds to my astonishment about it degrading. Usually vendors choose such a long warranty if they believe in their product - otherwise it's going to be expensive for them.

Regards
Dalai