Question Black screen freeze, requiring manual shutdown- PSU?

[mellow123]

Hey there,

for some weeks now i've got this problem where my PC freezes and black screens on me, requiring a manual shutdown.
It happens randomly, while i browse, play a game or just idling.
I've reinstalled Windows 10, checked my RAM with memtest86 and the Windows ram checking tool, those didn't report any errors.
I used OCCT to run GPU, CPU and PSU tests for a few hours, my PC didn't reboot during any of these, and temps are just fine (under 70°C for my graphics card, under 65°C for CPU).
I updated my BIOS to the latest version, turned off XMP for my RAM. I checked if my graphics card has the latest firmware installed, it does.
The CPU, GPU and RAM aren't overclocked.
I also cleared CMOS.
Nothing seems to help and i'm kinda lost on what to do apart from replacing parts and seeing if my PC doesn't reboot for a week as it could be fine for three days and then just reboot on the fourth.
My 12V Rail voltage (as taken from HWiNFO) reads 11.576V to 11.837V, could that be the culprit?

I've got a Corsair CS650M.

 

[jonnyguru]

PSU?

No.

 

[mellow123]

No.

Why not? I get a Kernel error saying that the shutdown happened due to unexpected loss of power. (Event ID 41 in the Event Viewer)

 

[jonnyguru]

Why not? I get a Kernel error saying that the shutdown happened due to unexpected loss of power. (Event ID 41 in the Event Viewer)

Which is what you get anytime Windows is not properly shut down for any reason.

You said yourself that you have to manually shut down the PC to get it to come back up again. That is LITERALLY an unexpected shut down.

https://docs.microsoft.com/en-us/wi...n: The system has rebooted,or by a Stop error.

 

[mellow123]

Which is what you get anytime Windows is not properly shut down for any reason.

Would you have any pointer as to what it could possibly be?

 

[jonnyguru]

Bad motherboard?

 

[mellow123]

Bad motherboard?

On the page you linked under Section 3 it says if the BugcheckCode is 0 and the PowerButtonTimestamp is 0 (which they both are) it's possibly a faulty PSU. As i checked the memory, didn't OC my Ryzen 5, and my PC doesn't overheat, it should be the PSU, no? Thank you for linking that.

 

[Fkort96]

Had the same problem once at a friend of mine. It was indeed the power supply. It usually came paired with full fan speed from the PSU. I suggest you borrow or buy another power supply just in case (atleast in NL where I live they have good 14 day return policies) . Try to run Prime95 and Furmark and use HWInfo to monitor the voltages.

 

[jonnyguru]

On the page you linked under Section 3 it says if the BugcheckCode is 0 and the PowerButtonTimestamp is 0 (which they both are) it's possibly a faulty PSU. As i checked the memory, didn't OC my Ryzen 5, and my PC doesn't overheat, it should be the PSU, no? Thank you for linking that.

Ok. Go ahead and change the PSU. Let us know if it doesn't work.

Again: If you initiate a power off/reset without going through START>SHUTDOWN>RESTART, you will ALWAYS get a Kernel 41.

 

[Karadjgne]

Black screens and freezing is almost always the gpu.

With that psu, you aren't going to shutdown half a rail, where the pc still has the power enough to show a frame (even if frozen) and not enough to make the gpu show the next frame. With a psu, it either works or it doesn't. If the 12v rail (that's what powers the gpu) gets nuts or hits a protective trip etc, the 12v rail would shut down entirely. The cpu is also powered by the 12v rail.

Meaning the psu isn't glitching or you'd loose all power to everything and pc would not require a manual shutdown. It works or doesn't, there's no decision to only work partially and only supply a certain amount of wattage.

The other psu, I'm betting had multiple 12v rails that were not tied together. You'll see that as 12v1 16A, 12v2 18A etc and listings for 12v1 cpu, 12v2 pcie, 12v3 motherboard etc. If one of those rails quits, doesn't affect the other 3.

The CS is Single rail.

I'd be looking at everything from gpu drivers to motherboard drivers, especially audio and Lan. That means using DDU (guru3d.com) to delete the old drivers in safe mode and going to the motherboard website for chipset driver updates.
Windows just had a major update, so that's a known cause of messing with gpus.

 

[mellow123]

Black screens and freezing is almost always the gpu.

With that psu, you aren't going to shutdown half a rail, where the pc still has the power enough to show a frame (even if frozen) and not enough to make the gpu show the next frame. With a psu, it either works or it doesn't. If the 12v rail (that's what powers the gpu) gets nuts or hits a protective trip etc, the 12v rail would shut down entirely. The cpu is also powered by the 12v rail.

Meaning the psu isn't glitching or you'd loose all power to everything and pc would not require a manual shutdown. It works or doesn't, there's no decision to only work partially and only supply a certain amount of wattage.

The other psu, I'm betting had multiple 12v rails that were not tied together. You'll see that as 12v1 16A, 12v2 18A etc and listings for 12v1 cpu, 12v2 pcie, 12v3 motherboard etc. If one of those rails quits, doesn't affect the other 3.

The CS is Single rail.

I'd be looking at everything from gpu drivers to motherboard drivers, especially audio and Lan. That means using DDU (guru3d.com) to delete the old drivers in safe mode and going to the motherboard website for chipset driver updates.
Windows just had a major update, so that's a known cause of messing with gpus.

Thank you, that makes sense.
This is a fresh install of Windows which i've installed the latest DCH drivers on, so i wouldn't need to use DDU, right?
Also, i have a steinberg audio interface with its own driver, so those audio drivers would be unnecessary, no?
I will install the Lan chipset drivers and see if that fixes anything.
Should i install the chipset drivers from Asus' or AMD's website? AMD's seems to be newer.
Also, Asus' driver package wants me to install some Radeon Software Adrenalin Edition before actually installing the chipset drivers (???).

 

[Karadjgne]

With Ryzen there's 2 types. Cpus and APU's. The cpus don't have integrated gpu, the APU does. The APU will require the Asus drivers, cpu doesn't.

You shouldn't need to add APU drivers to a cpu.

The cpu is going to process audio using directX and other motherboard components before it sends it externaly. So yes, it's not going to hurt to have updated audio files as that's the source, how the audio files are used/translated/rendered, comes even before a sound card.

Motherboard chipset drivers get from Asus. AMD is generic. Asus is tailored, rewritten, revamped specific to the chipsets on your particular motherboard.

You'd use Amd drivers on an Amd motherboard, not an Asus motherboard built to use Amd equipment. Same with bios.

 

[mellow123]

With Ryzen there's 2 types. Cpus and APU's. The cpus don't have integrated gpu, the APU does. The APU will require the Asus drivers, cpu doesn't.

You shouldn't need to add APU drivers to a cpu.

The cpu is going to process audio using directX and other motherboard components before it sends it externaly. So yes, it's not going to hurt to have updated audio files as that's the source, how the audio files are used/translated/rendered, comes even before a sound card.

Motherboard chipset drivers get from Asus. AMD is generic. Asus is tailored, rewritten, revamped specific to the chipsets on your particular motherboard.

You'd use Amd drivers on an Amd motherboard, not an Asus motherboard built to use Amd equipment. Same with bios.

I see, i installed the chipset , the Lan and Audio drivers. Sadly, the crash occured again just now. The PC rebooted (without me having to press the power button) and there are two new errors in the event viewer, listed 5 seconds after the Kernel Power Error with ID 41.

Error 1:
WHEA-Logger (Event ID 18)
A fatal hardware error has occurred.

Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Translation Lookaside Buffer Error
Processor APIC ID: 1

Error 2:
WHEA-Logger (Event ID 18)
A fatal hardware error has occurred.

Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 10

The crash happened just after finishing a game of Valorant.
I could also post the HWiNFO logs but i didn't see anything alarming there on first glance.
I've got a Fedora installation as well on this drive, i remember the error occuring there as well.

 

[Karadjgne]

https://social.technet.microsoft.co...icrosoft-windows-whea-logger.aspx#Explanation

Psu is a suspect, but only because there a slew of uber-cheap knockoffs with 20 year old platforms that cost $3 in parts and sell for $30.

Id be thinking more along the lines of a cpu error, especially with cache errors, but could be as simple as corrupted data. Possibly a motherboard failure, VRM going bad or overheating, and not supplying correct voltages when the cpu demands it
https://www.windowscentral.com/how-use-dism-command-line-utility-repair-windows-10-image?amp
Can check those 2 to see windows health.

Also run HWInfo (sensors only) and set it to log when gaming. You can check the temps, voltages etc after.

 

[mellow123]

https://social.technet.microsoft.co...icrosoft-windows-whea-logger.aspx#Explanation

Psu is a suspect, but only because there a slew of uber-cheap knockoffs with 20 year old platforms that cost $3 in parts and sell for $30.

Id be thinking more along the lines of a cpu error, especially with cache errors, but could be as simple as corrupted data. Possibly a motherboard failure, VRM going bad or overheating, and not supplying correct voltages when the cpu demands it
https://www.windowscentral.com/how-use-dism-command-line-utility-repair-windows-10-image?amp
Can check those 2 to see windows health.

Also run HWInfo (sensors only) and set it to log when gaming. You can check the temps, voltages etc after.

But is it likely that the CPU would just fail like this after 3-4 years of working great? It never overheated or anything, i've got an aftermarket cooler (Hyper 212 Evo) installed.
I did plenty of stress tests using OCCT for CPU, GPU and PSU, all of those returned no errors.
I used DISM, came back with no errors.

Here is the HWiNFO log of when my PC crashed yesterday: https://we.tl/t-7b2XcG2FNS
(last logged values at the bottom).
If i had to guess i would say probably psu, mobo or some weird windows update messing with the chipset driver or something?

 

[infirmusetiam]

i had same problem freeze randomly on win 10 since 3 years , i just installed a custom os of win 10 thats all (iso from team os ) really bro

 

[Karadjgne]

Win10CE is unlike any prior OS. It's a priority OS. It like to use its own drivers over others just based on version numbers and timestamps. Which is a bogus problem way. When it first dropped and with every successive major update, it gives fits to many drivers simply because it's newer. Newer doesn't mean better when the older file is tailored for use. It's a big issue for things like ASmedia Sata controllers which are tailored specifically for your motherboard chipset to get maximum performance, and windows steps in and uses a 'newer' version of Sata controllers that's totally generic, so gets worse performance. Start applying that to gpu drivers or audio drivers or Lan drivers and things gets nuts in a hurry.

It was so bad that MSI (bless their corporate hearts) went back and updated some files for my Z77 motherboard which hadn't been touched since mid-year 2013, just to make them Win10CE compliant. Worked fine with 7 or original 10, but not with Creators Edition or it's successor.

 

[Karadjgne]

If I'm reading that log right, you've got cores hitting in the 90's at times, specifically core#5. You'd be seeing an aggregate temp, maybe 60's for the cpu, but that's 5 cores at mid 50's and 1 core at 90.

If I read that right.

Which could be fluff, dried paste, air pocket, paste not spread right (Ryzen 3 series is not a center located monolithic block of cores, it's 2-3 chiplets spread out near the edges, so many times the 'pea method' isn't that good)

Might need to check the paste, clean and redo, and I'd use the 'credit card' swipe and cover all corners method to make doubly sure the full IHS is covered well.

 

[mellow123]

If I'm reading that log right, you've got cores hitting in the 90's at times, specifically core#5. You'd be seeing an aggregate temp, maybe 60's for the cpu, but that's 5 cores at mid 50's and 1 core at 90.

If I read that right.

Which could be fluff, dried paste, air pocket, paste not spread right (Ryzen 3 series is not a center located monolithic block of cores, it's 2-3 chiplets spread out near the edges, so many times the 'pea method' isn't that good)

Might need to check the paste, clean and redo, and I'd use the 'credit card' swipe and cover all corners method to make doubly sure the full IHS is covered well.

Can you tell me which column that is? Can't seem to find it.

 

[Karadjgne]

AJ-AW I think, hard to read on mobile. But you got cores averaging @ 20-30's and one that consistently skyrockets. Ryzen 3000 uses a preferred core thing, which ever core it decides is the better core is what predominantly gets used, until it gets a little worn and then it'll switch to another preferred core. It's the new standard put into affect with 3000 series and 10th Gen Intel to try and get best performance consistently on minimum amount of core use. Same theory as rotating tires, prolong the life of all by not wearing one out prematurely.

So seeing higher temps on one to three cores is to be expected, but 90's+ is high.

 

[mellow123]

AJ-AW I think, hard to read on mobile. But you got cores averaging @ 20-30's and one that consistently skyrockets. Ryzen 3000 uses a preferred core thing, which ever core it decides is the better core is what predominantly gets used, until it gets a little worn and then it'll switch to another preferred core. It's the new standard put into affect with 3000 series and 10th Gen Intel to try and get best performance consistently on minimum amount of core use. Same theory as rotating tires, prolong the life of all by not wearing one out prematurely.

So seeing higher temps on one to three cores is to be expected, but 90's+ is high.

Oh, i see what you mean. I think that is just the thread usage though as the unit shows it's in %.

 

[Karadjgne]

Ahh gotcha. Excel and mobiles aren't always kosher lol
C5T0 and C5T1 would lead to believe temp, not usage.

 

[mellow123]

https://imgur.com/a/AyZBc0K

View: https://imgur.com/a/AyZBc0K


Are you referring to those two? Those seem to be usage of thread 1 and 2.