help i never seen this msi behavior before

[lordallen14]

help me diagnose this msi build, pc is about 2years old now

pc randomly shutting off this past month just from watching youtube or gaming and after that when we try to power back on, it just power cycles for 7 to 10 seconds, red and orange led, then turns off completely.

but doing a full cmos power drain reset(remove battery, shorting jbat jumper,pressing,cmos button) suddenly makes pc 100% working fine again
so basically i have to cmos reset every single time this shutoff happens.

spec: MSI mpg b650 carbon wifi,ryzen 9 7950x,5070ti,32gb ddr5 expo off, pbo off, (bios stock settings), 850w seasonic gold atx 3.1, windows10 pro, bios and chipset updated to latest

tried replacing bnew cmos battery, and already RMA the ram i got bnew ram replacement, psu is quite new about 4months after i upgraded to 5070ti, reseated cpu aio mount, cleaned ram slot
any ideas what could be causing this?

 

[citay]

tried replacing bnew cmos battery,

The battery lasts 5-10 years, this can't be it, and a dead battery will not cause a sudden shutdown. At least this doesn't cost much and isn't much effort, but in such cases it will very likely be useless.

and already RMA the ram i got bnew ram replacement,

Never RMA something without having done proper troubleshooting to narrow it down to that specific component as the culprit. Otherwise, you risk it being a waste of time and resources. When you suspect the RAM, test with one module of known good RAM. But bad RAM is very unlikely to cause a sudden shutdown.

The most likely causes for sudden shutdowns are things like CPU overheating and reaching the shutoff temperature, from a failed water cooler for example, like a dead pump or a clogged liquid circuit. The PSU is another possibility. So one of the most important things to know here is the cooler model. Have you looked at the temperatures? For example, what is the CPU temperature when you just let it sit in the BIOS for a bit? Speaking of which, is the BIOS up-to-date?

 

[plutomate]

Shutting down directly or having any sort of error code? Have you updated windows lately? What if not doing any game, browsing webpage maybe? Does the system still turn off?

 

[Jarous1]

This is the absolute dumbest question: Have you plugged in the graphics card's external power connector/s? Since you mentioned that you replaced GPU...
The computer usually works fine without it (even though the GPU drivers should report that the connector(s) is not connected), but when it runs something more graphics-intensive (even YouTube), the system usually crashes.

 

[lordallen14]

no signs of cpu overheating. it idles around 30c, gaming 57c-60c, cinebench r23 full throttle reaches 86c.

tpm errors also keep spamming windows event viewer. i tried turning tpm off and on, but nothing changed. could a faulty tpm chip somehow short or affect the whole system? could this be a bad motherboard? i wish i could rma the motherboard, but i dont have a spare to test with. im worried about rma'ing the wrong part again since i already rma'd the ram thinking it was the culprit, but it turned out not to be the problem.

 

[citay]

Ok, the temperatures sound like the AIO is working. Otherwise they would be way higher even in idle.

Does the PC always shut down suddenly after a certain amount of time, and that time is always identical? Because we had something like that recently, see here, which was related to an issue with the Management Engine (which only Intel-based systems have, and you have an AMD-based system). But if your PC always shuts down after exactly 30 minutes for example, then it could be something along these lines, a similar mechanism.

 

[ericj159802e9]

pc randomly shutting off this past month just from watching youtube or gaming

This entices me to also bring the GPU into focus. You already tested with Cinebench. Could you test CPU and GPU with OCCT just to have a second opinion?

 

[lordallen14]

i ran furmark for 30 minutes with no crashes or shutdowns. max temp was only around 68c. it doesn’t seem to be specifically related to youtube or gpu load anymore. it feels pretty random and can even happen when the pc is just sitting idle.

i haven’t had any shutdowns for the past 2 days, but it still feels like a 50/50 chance that it could happen again.
what confuses me is that if the gpu were faulty, shouldn’t the pc still post? the debug leds show red for cpu and orange for dram, which makes me wonder if something is failing during the power-on self-test (post) rather than the gpu itself?

 

[citay]

If the shutdowns are truly completely random, and now they don't even happen anymore, then troubleshooting is difficult indeed. As mentioned, do not RMA anything else for now. An RMA should be the end result of successful troubleshooting to identify the culprit.

The troubleshooting would usually involve anything you can check without too much hassle, like is the BIOS up to date, Clear CMOS, re-plug the power cables on the board/GPU and PSU side in case there is a loose connection somewhere, making sure the temperatures are good in all scenarios. A lot of what you did until now. Then maybe even taking out the CPU and checking the socket pins for any bent pins, because if there are bent pins, they can cause very weird symptoms sometimes.

Also, you can leave out anything for a test that is not strictly required. Use one RAM module only in slot A2, leave out the GPU and use the iGPU of the CPU. Of course, with the RAM having been RMA'd, that's completely off the radar now, but just saying how it's done in principle. And if all this was not helping, you would have to test with known good (borrowed) components, for example a known good PSU of decent quality, to rule that out also.

Now, about the CPU+DRAM LEDs, yes, that could point to the board having a problem. Sometimes the AM5 600-series boards in particular can develop these odd issues where CPU+DRAM LEDs appear one day. But usually it's even when first trying to boot it. It's more rare to have it shut down suddenly, and then show the LEDs.

Anyway, the problem now is that the shutdowns happen sporadically, if at all, so it becomes difficult to troubleshoot. Because you'd have to wait several days with each thing you try, to know wether it helped or not.