Question PC locking up randomly, only shutting down the pc helps

[CrytoxXD]

Hi all,

in Dec 2021 I built my own PC. After maybe a month or so, it started randomly locking up and only a hard-shutdown would unlock it.

What I mean with locking up is:
My PC completly freezes. Neither my mouse nor my keyboard can do anything. My microphone also stops working (using an external RODE with AI-1 interface) and my monitors stop moving (but they don't shutdown)
Also the sound of my PC keeps playing. I hear everything that is going on in the background but can't do anything.
Only holding down the power button will actually "free" my pc.

Hardware:
Mainboard: MSI MPG X570 Gaming Edge Wifi (MS-7C37)
CPU: AMD Ryzen 7 3800X
GPU: AMD Radeon RX 6700 XT (To be exact: XFX Speedster MERC 319 Radeon RX 6700 XT Black Gaming 12GB)
PSU: be quiet! PURE POWER 11 (700W)
RAM: 4x8GB G.Skill Aegis DDR4-3200 DIMM CL16
Drives:

• 2x 1TB Samsung SSD 980 M.2 PCIe 3.0 x4
• 1x 1TB TOSHIBA HDWD110 HDD

Ethernet-Card: Delock PCI Express LAN89564

OS - Windows 11 Pro (22H2) (22621.1105)

All drivers were last updated Dec 2022 as well as the BIOS.
What I did so far:

• I've run Memtest86 two times, both times without errors.
• I've reinstalled the OS and every driver multiple times
• I've tested all RAM-Sticks one at a time
• I've replaced my Mainboard with the same model but fresh from the shop
• I've tried to only use 1 or 2 monitors of my 3 monitor setup
• I've tried underclocking / overclocking each component
• I've tried using another mouse / keyboard
• I've tried using another CPU

It also doesn't appear to be overheating... the max. temp of the GPU is 75°C (min. at 27°C) and the CPU peaks at 60°C (min at 30°C).
It also only really locks up playing Games, but nothing high-class, mostly games like LOL or Valorant do the trick.

Nothing really helped and as of now, I'm out of ideas what to test anymore or even where to look for clues.
If any information is missing please tell me and I'm gonna post it asap. I am gratefull for any ideas or help solving this.

(Execuse my english, not my first language)

 

[Fix_that_Glitch]

Was the 4x8 ram all a single package, or was it two 2 x8 ?

 

[CrytoxXD]

Was the 4x8 ram all a single package, or was it two 2 x8 ?

2 Kits with 2x8GB

 

[Fix_that_Glitch]

You will have to look at the serial numbers on the memory sticks to match them in pairs. When you sort that out, try using only one pair at a time that match. See if the problem persists.

 

[CrytoxXD]

You will have to look at the serial numbers on the memory sticks to match them in pairs. When you sort that out, try using only one pair at a time that match. See if the problem persists.

Mh okay. Good idea. But just out of curiosity... why would the PC freeze because of that? And wouldn't have Memtest86 found that and marked it as an error or so?

 

[Fix_that_Glitch]

Mh okay. Good idea. But just out of curiosity... why would the PC freeze because of that? And wouldn't have Memtest86 found that and marked it as an error or so?

No memtest tests your ram sticks. It may not pick up that your sticks don't play well together. This is just to rule out your ram is the problem so you can move on to other troubleshooting ideas.

 

[Fix_that_Glitch]

Actually I did not see that you tried ram sticks one at a time os this would just be repeating the same test. Sorry.

 

[Fix_that_Glitch]

In event viewer, under computer management, are there any critical error reports?

 

[BlaUno]

Just to be sure, do you have 2x 8pin cables plugged in in your GPU, or is it daisy chained, in other words 1x cable with 2x8pin connectors on it? You need 2x separate cables from the psu to your gpu, ofc. Other than that, it could be the driver (possibly struggling with some remains of an older driver), in that case it might be worthy taking a look at DDU (Display Driver Uninstaller).
I've had similar problems with a garbage PSU but in that case it would either straight up crash or you hear the sound stuck in a robotic loop, which doesn't seem to be the case with you, even tho 700W kinda seems to me to be on the edge (from what i've read).

 

[CrytoxXD]

No memtest tests your ram sticks. It may not pick up that your sticks don't play well together. This is just to rule out your ram is the problem so you can move on to other troubleshooting ideas.

Oh okay. Thanks for the explanation.

Actually I did not see that you tried ram sticks one at a time os this would just be repeating the same test. Sorry.

I just tested it again just in case.

In event viewer, under computer management, are there any critical error reports?

There are a few, yes. All from Kernel-Power with ID 41. All saying: "The System was restarted without being shutdown properly. [...]"
They could be from the lockups, because I had to force a shutdown everytime.

Just to be sure, do you have 2x 8pin cables plugged in in your GPU, or is it daisy chained, in other words 1x cable with 2x8pin connectors on it? You need 2x separate cables from the psu to your gpu, ofc. Other than that, it could be the driver (possibly struggling with some remains of an older driver), in that case it might be worthy taking a look at DDU (Display Driver Uninstaller).
I've had similar problems with a garbage PSU but in that case it would either straight up crash or you hear the sound stuck in a robotic loop, which doesn't seem to be the case with you, even tho 700W kinda seems to me to be on the edge (from what i've read).

It's a daisy chained cable. I didn't think that would make a difference tbh... and 700W is indeed on the edge, but after some reading and some talking to some work-buddies (I work in IT, but in the software sector) that the entire usage should peek at ~690W so ~10W padding...
I think I'm gonna change the cables to 2 separate from the PSU to the GPU and maybe I'll uprade to a 750W PSU if that doesn't help.
And for driver remains... I actually DDU before so I don't think there should be any. But I'll keep it in mind if the changing of the cables and change of PSU doesn't do anything.

 

[letmepicyou]

Oh okay. Thanks for the explanation.


I just tested it again just in case.


There are a few, yes. All from Kernel-Power with ID 41. All saying: "The System was restarted without being shutdown properly. [...]"
They could be from the lockups, because I had to force a shutdown everytime.



It's a daisy chained cable. I didn't think that would make a difference tbh... and 700W is indeed on the edge, but after some reading and some talking to some work-buddies (I work in IT, but in the software sector) that the entire usage should peek at ~690W so ~10W padding...
I think I'm gonna change the cables to 2 separate from the PSU to the GPU and maybe I'll uprade to a 750W PSU if that doesn't help.
And for driver remains... I actually DDU before so I don't think there should be any. But I'll keep it in mind if the changing of the cables and change of PSU doesn't do anything.

If your system uses 690 watts at peak draw, then 750 watts is NOT ENOUGH. The closer you run your power supply to its maximum output, 1: the hotter it runs 2: the less efficient (and thus more expensive) it is to run 3: the less time it lasts.

As a rule of thumb, I won't build a system unless I have at least 20% overhead on the power supply. MINIMUM.
You would want to put a QUALITY 850 watt power supply in that rig.

 

[BlaUno]

It's a daisy chained cable. I didn't think that would make a difference tbh...

To quote a reddit post i've found, which aligns with other stuff i've read:

„RX 6700xt can pull upto 230 watts max, 75 watts from the motherboard, rest from the cables, 8 pin pcie cable is rated for 150w. 1 cable will do the trick, but 1x 8 pin and 1x 6+2 pin is optimal“.

Then there are others who claim that „most higher end GPUs don't really draw much power from the MB, some draw even less than 10W and rely on the cables.“...

Bottom line, plugging two separate cables sounds like the safer option.

 

[CrytoxXD]

If your system uses 690 watts at peak draw, then 750 watts is NOT ENOUGH. The closer you run your power supply to its maximum output, 1: the hotter it runs 2: the less efficient (and thus more expensive) it is to run 3: the less time it lasts.

As a rule of thumb, I won't build a system unless I have at least 20% overhead on the power supply. MINIMUM.
You would want to put a QUALITY 850 watt power supply in that rig.

Mh... valid point. I'll just buy an 850W PSU just to rule that out. Thanks for the hint and thanks for that rule of thumb, gotta remember that one!

To quote a reddit post i've found, which aligns with other stuff i've read:

„RX 6700xt can pull upto 230 watts max, 75 watts from the motherboard, rest from the cables, 8 pin pcie cable is rated for 150w. 1 cable will do the trick, but 1x 8 pin and 1x 6+2 pin is optimal“.

Then there are others who claim that „most higher end GPUs don't really draw much power from the MB, some draw even less than 10W and rely on the cables.“...

Bottom line, plugging two separate cables sounds like the safer option.

Interesting... Yeah I think I'm just gonna plug the other cable in. Maybe the problems really just come from way to less power. I'll see and post asap once I could either confirm or deny it.

But already thanks for that amount of replies, guys and gals!

 

[Fix_that_Glitch]

I know you said you installed drivers over and over again, does that include updating the bios?

 

[DSzymborski]

And remember, this particular Pure Power 11 CM isn't one of their high-end ones; it's only specced for 672W of +12V power. It was an older one, too as they've long since phased those out for the Pure Power 11 FM PSUs.

 

[Fix_that_Glitch]

Oh okay. Thanks for the explanation.


I just tested it again just in case.


There are a few, yes. All from Kernel-Power with ID 41. All saying: "The System was restarted without being shutdown properly. [...]"
They could be from the lockups, because I had to force a shutdown everytime.



It's a daisy chained cable. I didn't think that would make a difference tbh... and 700W is indeed on the edge, but after some reading and some talking to some work-buddies (I work in IT, but in the software sector) that the entire usage should peek at ~690W so ~10W padding...
I think I'm gonna change the cables to 2 separate from the PSU to the GPU and maybe I'll uprade to a 750W PSU if that doesn't help.
And for driver remains... I actually DDU before so I don't think there should be any. But I'll keep it in mind if the changing of the cables and change of PSU doesn't do anything.

The max output is actaully 672 watts on the 12v rail/s so you are even providing less than 700 watts to the system.

 

[BlaUno]

@CrytoxXD On the topic of drivers and DDU. Check out this video from this guy Jayztwocents. Basically explains that Windows "Fast startup" and "Automatic installer" can interfere with DDU, so it's advised to turn them off before using it. As he says himself, he did a driver wipe with DDU just before the video but just because he forgot to disable these, installing the new driver again resulted in issues. So he had to redo the whole process, except this time he recorded it so that's what the video is:

View: https://youtu.be/F8OLhUAPDq0

 

[CrytoxXD]

I know you said you installed drivers over and over again, does that include updating the bios?

Yes. Most times I looked every month if there is a stable patch and installed.

And remember, this particular Pure Power 11 CM isn't one of their high-end ones; it's only specced for 672W of +12V power. It was an older one, too as they've long since phased those out for the Pure Power 11 FM PSUs.

Oh boy... I didn't knew that fact... Just out of curiosity where can you find information like that? Directly on the website of the product?

The max output is actaully 672 watts on the 12v rail/s so you are even providing less than 700 watts to the system.

That could explain really much actually...

@CrytoxXD On the topic of drivers and DDU. Check out this video from this guy Jayztwocents. Basically explains that Windows "Fast startup" and "Automatic installer" can interfere with DDU, so it's advised to turn them off before using it. As he says himself, he did a driver wipe with DDU just before the video but just because he forgot to disable these, installing the new driver again resulted in issues. So he had to redo the whole process, except this time he recorded it so that's what the video is:

View: https://youtu.be/F8OLhUAPDq0

Uh nice! Thanks for that vid. I think I'm gonna do a complete driver wipe after switching to a better PSU, just to rule out faulty drivers.

 

[CrytoxXD]

Hello fellas,

an Update because of... reasons:
Friday I installed a newly bought PSU (bequiet! Dark Power 12 850W) and plugged two seperate cables into the GPU. So power should not be an issue anymore in my rig, I think.

I also clean reinstalled both AMD Drivers (GPU and Chipset) (Again thanks @BlaUno for the awesome video!) and checked if there is a new BIOS update available (nothing yet)

Although it didn't lock up for quite some time, it eventually did 2 hours ago.
It occured while playing a round of LOL (League of Legends). In the 4 round of the day. I had my taskmanager open whilst it happend so some details:

CPU Usage: 23%
GPU Usage: 46% with temp of 43°C
RAM Usage: 11,3GB / 31,9GB so about 33%

Also worth telling is that I actively use 3 monitors with 240hz. The PC freezed with LOL on my main monitor, Chrome open with an YT Video (Music Mix) on the second and Discord and Taskmanager on the third being split in the middle.

Again, I'm asking for help because changing the PSU doesn't seem to have helped. Every bit of helped is appreciated.

 

[BlaUno]

...Chrome open with an YT Video (Music Mix) on the second and Discord and Taskmanager on the third being split in the middle...

I've found this, here, they talk about very similar if not the same exact issue.

„ Got same issue on 6750xt.. It's been ok last few days after I disabled hardware acceleration on chrome and discord.“
And then someone else responds:
„ lmao this is such a hilarious fix. This has worked for me for the last 24hours“.

On that note, i've seen lots of forum posts where turning off hardware acceleration is one of the main advises (afaik it wasn't mentioned here so far). Also, very recently there was news that Discord seriously butchers performance on nVidia cards, so based on that, i wouldn't be surprised if it crashes AMD cards as well.

Another thing i would try if that doesn't work is updating vBios, even tho i know nothing about it. I'm assuming it's the GPU firmware or something. That same youtuber i linked before had another video where he updated the motherboard BIOS to the latest version in order to use a new CPU and that unexpectedly stopped the GPU from working. He fixed it (after some troubleshooting) by updating the v(ideo?)bios, after coming to a conclusion that the newest mobo BIOS jumped so far ahead that it rendered the GPU bios outdated and incompatible at that point. At least that's how i understood it. The video itself is irrelevant for this, because he didn't actually show HOW to do it, but i'm sure you'll find instructions online, if you think this is relevant at all.

 

[CrytoxXD]

I've found this, here, they talk about very similar if not the same exact issue.

„ Got same issue on 6750xt.. It's been ok last few days after I disabled hardware acceleration on chrome and discord.“
And then someone else responds:
„ lmao this is such a hilarious fix. This has worked for me for the last 24hours“.

On that note, i've seen lots of forum posts where turning off hardware acceleration is one of the main advises (afaik it wasn't mentioned here so far). Also, very recently there was news that Discord seriously butchers performance on nVidia cards, so based on that, i wouldn't be surprised if it crashes AMD cards as well.

Another thing i would try if that doesn't work is updating vBios, even tho i know nothing about it. I'm assuming it's the GPU firmware or something. That same youtuber i linked before had another video where he updated the motherboard BIOS to the latest version in order to use a new CPU and that unexpectedly stopped the GPU from working. He fixed it (after some troubleshooting) by updating the v(ideo?)bios, after coming to a conclusion that the newest mobo BIOS jumped so far ahead that it rendered the GPU bios outdated and incompatible at that point. At least that's how i understood it. The video itself is irrelevant for this, because he didn't actually show HOW to do it, but i'm sure you'll find instructions online, if you think this is relevant at all.

Thanks for the reply! Yeah i disabled it on both now, and thus far everything seems to be working. I'm gonna monitor it for a while.
And that thing with vBios... I actually already thought about that part but I was a bit hesitant because updating it can result in some major damage. But definetly thanks for the heads-up I'm gonna keep it in mind.