Question IMPOSSIBLE Shutdown problem

[Leonardo Weiss]

Hi,

Since I bought my new pc last year I've been dealing with some shutdown problems, but I still don't know how could I fix this problem.

The PC basically turns off in random occasions... such as:
-4K Gaming
-Rendering 3D stuff

I have already tried A LOT of possible fixes, but it just doesn't works, and it's so frustraing.

Here's everything I've tried:
-Formatting PC
-Cleaning the dust
-Changing CPU thermal paste
-Changing RAM slots
-Turning Off X.M.P RAM profiles
-Checking PSU cables
-Changing PSU
-Undervolting CPU
-Tried RAM sticks from friends
-Changed cpu cooling curves
-Ran the pc with the case opened
-Stress Tests (and got no shutdowns[!?!?!?])
-Reset Bios Settings
-Updated BIOS
-Updated Drivers
-Changed PC power plan
-Malware Check
...
It should be more, but I don't remember now...

And I still have no clues about what is going on.

It would be obvious to think that something is damaged, but none of the components have ever gave any indication of malfunction.

CPU: Ryzen 7 3700X (Stock Cooler) - Running Undervolt at 4100Mhz and 1.150V
Temperatures around 60C when gaming.

GPU: AMD 5700XT (MSI) - Temp around 80C-90C gaming, Performance exactly how it should be.

RAM: 2x16GB XPG Spectrix - I don't use X.M.P configs because it was overheating my CPU. This RAM is not actually 100% compatible with my MOBO (it doesn't show this memory card name on the compatible memories list from the mobo) but searching online, people always say that this doesn't matter.

MOBO: Aorus B450M Elite V2 - I think I broke one RGB socket (that's all about the mobo)

SSD: SanDisk 240GB
HDD: Seagate 2TB 7200RPM

PSU: EVGA 600W 80 Plus Bronze when I first started getting these shutdowns, but then I changed to a XPG 850W 80 Plus Gold.

I'm thinking about trying different electrical outlets, reinstalling my previous BIOS version and Running Stress, Benchmarks and tests all over again.

I'm completely lost, and so frustrated! I've spent my life savings on this PC, and it just literally doesn't works properly!

Hope someone have the golden answer, but even if you don't, but you know something it should help, please help me!

Thanks in advance.

 

[4745454b]

Undervolt the CPU? When you are having issues you need to restore everything to default. No XMP, no OC, no changing the voltage on anything. Restore EVERYTHING to how it should be. Does it work?

Anything in the event viewer? Knowing what is having issues will help figure out what to do.

 

[Leonardo Weiss]

Hey, thanks for replying!

When I do default configs the PC shuts down even faster!
You may think that the undervolting config is not stable and not providing enough power... But last time I've checked it, it was not. But I will give it a try and check it once again.

But the thing is, when I'm not doing any undervolt config, the PC shuts down even faster, about 4-5 minutes playing GTAV in 4k for example, 'cause the CPU temps goes all over 75 degrees. But when doing the undervolt, it goes 60º max, and it takes half an hour to shuts down.

But yeah, maybe for some reason is no more stable... I'll check it later.

There's also been a long-time since I last checked the event viewer. Last time (a few months ago) there was an error in there every time I got a shutdown, don't remember the exact error code, But I remember searching about it, and there was nothing useful to find online.
I will also check it once again when I get home today.

Once again, thanks for replying!

 

[4745454b]

The error code might have been a not helpful "unexpected loss of power". Duh. But why PC?

Undervolting helps. That would be a temp issue. Undervolting helps the temps stay low so it lasts longer, but if the CPU needs more volts then it can throw an error. Look at your cooling. Why is that not working? Dusty heatsinks? Improper TIM/paste applied? Fan not spinning fast enough? I'm not there so I can't point to X and say that's it.

You might also want to check for cooling on your motherboard. VRMs can get really hot.

 

[Leonardo Weiss]

I GOT SOME BIG NEWS!
But first:
At the Event Viewer, I've found two different events happening.
Event ID 6008 and ID 41

ID 6008 is the most frequent one, while ID 41 just happens once in a while.
After searching about it online, I went through some tutorials that made me change some power settings and stuff but after trying the GTA again, nothing has changed, the PC turned off again.

I've been doing some stress tests on my CPU, in a way to get a shutdown and adjust my undervolting setting, but got no shutdowns.

I'm using a Brand New Wraith Prism Cooler with no dust and paste properly applied.
I've seen a lot of people complaining about CPU temperatures from Ryzen 3000 series, I think undervolting it's a common thing for Ryzen 3700x owners.

And VRMs, don't seem to be that much hot. Temps are between 40-75 during stress tests.

Now, the big updates!

As a way to do the stress tests, I've downloaded a lot of monitoring and stress software.
And after doing a lot of CPU stress tests, I tried doing one GPU stress test.

5 minutes was enough to get a complete shutdown!

So the problem is GPU-related!?
But the temperatures are not getting higher than 85ºC, what is the problem then?

Power? But I got a brand new PSU, it can't be power.

Maybe the GPU is damaged? But if that were the case, why I'm not getting artifacts or performance issues?

Could it be related to my house's electricity?

 

[4745454b]

Which EVGA 600? You said you have an EVGA 600W bronze, and they aren't known to be great. It's most likely a group regulated design, with 432W on the 12V rail. It's very possible your 5700XT is spiking above it's rated TDP.

https://www.tomshardware.com/reviews/aorus_rx_5700_xt-8g/4

When using Furmark to look at power consumption, we see all the cards used hit the power limit. In this case, we observed power limits for the Aorus 5700 XT around 225W

If I'm right, the 5700XT is spiking above the 225W, and when you add the CPU and the rest of your system is going above what your PSU can handle. Getting a better PSU would solve this issue.

 

[Karadjgne]

He replaced the Evga with an XPG 850w Gold. That should be enough to handle the 5700xt spikes. It's not a bad unit.

Related to house power? Possibly. Figure your pc is pulling @ 5A from the wall. There's every other plug inline before that plug, there's connections to the breaker and panel. Just takes 1 lousy connection in any of those and you get overheating and certainly loss of connection, which kills power longer than the holdup time (20-30ms) of the psu, which then suddenly shuts off. Seems to only happen when under heavy loads.

You'd need to check all plugs starting with the pc, then chase them back to the panel, then check the breaker and neutral bar connections.

 

[Leonardo Weiss]

Exactly! I have replaced the EVGA with an XPG 850W Gold.

Testing the GPU again I got no shutdowns! I usually get a shutdown every 5min stressing or gaming, but this time nothing has happened for HOURS!
Even with 1 hour of GPU stress test and 1 hour of playing GTA nothing has happened. I got so confused and start searching online about it, after one hour of not pushing the PC, I decided to try again playing GTA.

And you know what? 5min shutdown again.
It just makes ANY sense.

I'll check all plugs later, as you said.

Something I've noticed during the last shutdown is some different fan noise... I'm not 100% sure because it's very subtle but I think the fans sound like they're wiggling, a few minutes before the shutdown, does that say something to you?
I'm not sure, but I think the sound comes from the GPU fans.

 

[4745454b]

Ah. Sorry, I missed that.

Can you try the 5700XT in another system? A friend perhaps? Try the PC setup on a different circuit? Not just a different wall plug, but on an entire different circuit in the breaker box. I'd also try to make sure any high draw devices like AC or dryer are off. If that works, perhaps turn one or both on? Maybe something else is turning on causing the issue. That's why it works sometimes but fails others?

 

[Karadjgne]

If the fans sound funky, just before shutdown, immediately look at the fans. If the fan over the gpu VRM's is slowing or freezing up, that'll do it.

And they won't show on anything but HWInfo64 as it's equipped to register more than one location for gpu temps. If the VRM's are overheating, they'll shutdown the gpu.

 

[Leonardo Weiss]

Since yesterday was Sunday, my home was completely quiet, the only devices connected to electricity were the lamps, a television, and my pc setup. Even though, I got something like 5-6 shutdowns.
On busy days when everything is turned on, the same issue happens with a LOT of devices connected to power.
So I don't think it comes from high draw devices.

I'll try testing the PC on my aunt's house to see if the power goes differently.

About trying the GPU in another system, I've been thinking to call a friend to my house and ask him to bring his PC, so I would be able to do a LOT of experiences.

About the fans:
It's not like it's slowing down, so it's hard to tell if any fans are performing wrong, It sounds like the speed is going up and down over and over again, so it's not a static noise but a strange "up and down" noise.
But yeah, I'll try to notice visually next time.

I'll check VRMs temps later, I saw many people with the same issue, and VRM temps seem to be a thing in many of those cases.