"New" PC crashes: No video signal,no Bluescreen, no Event Viewer entry

itsme1 said:

Make a small partition, install windows on it with the latest updates. If you have no more problems, install the drivers for the computer.

I've made a small 70 GB partition (better to have some space just in case).
I've downloaded the MediaCreationTool22H2 ( https://www.microsoft.com/en-us/soft...load/windows10 or https://www.microsoft.com/de-de/soft...load/windows10 ), and am gonna use a USB stick for this.


Just to be clear: Of course I don't want this to mess with the already existing OS, or delete any of my files.
Will the installer give me the option to choose where to install Windows 10, and it'll leave the other partitions as they are?
Is there a specific section in the installation I should pay special attention to, or anything else I should know about? At which point do I choose the partition to install Windows to?

Megahertz said:

@HedgeToaster, did you see may post #5 above?

Yeah, I saw it right after posting my reply to the others. Sorry I didn't get around to responding sooner.

Megahertz said:

It could be so many things. My guess is the power supply or MB.
- To eliminate a windows problem, do a repair install.

As @itsme1 suggested, I'm trying to install a fresh copy of Windows 10 on a new 70 GB partition, and see if the crashes persist for the (more or less) reproducible test case described in the original post.

This way I can definitely decide if it's a hardware or software issue, and act accordingly. If it's just a software issue, then there's no rush to fix it within the warranty period.

I'll get around to some of your other suggestions (I'm still looking through the options in OpenHardwareMonitor), but I'll try this first.

NTN said:

OCCT, a free CPU/GPU/PSU stress testing tool. OCCT stress tests your CPU, GPU, and motherboard, and in the process, can discover any faults with your power supply unit.
How to Fix a Windows 10 Kernel Power Error in 5 Easy Steps

Is using OCCT to stress test the system safe?

I've read about some people nearly "melting" their GPUs with it, or that a stress test like that can damage some PSUs. Especially if you run it for an hour as it is recommended.

I'll get around to the rest once I've done testing with that separate Windows 10 installation, as well as with OCCT if that's safe to use.

NTN said:

No clue what kind of Power State you got. (powercfg /a in CMD(adm))

Here's the result of that:

The following sleep states are available on this system:
Standby (S3)
Hibernate
Hybrid Sleep
Fast Startup

The following sleep states are not available on this system:
Standby (S1)
The system firmware does not support this standby state.

Standby (S2)
The system firmware does not support this standby state.

Standby (S0 Low Power Idle)
The system firmware does not support this standby state.

Malneb said:

The difference is that blue screen could effectively stay indefinitely without any harm to the computer [...] Black screen to reboot means fatal error in terms of something has happened and the computer thinks it needs to cycle to save itself from damage hence why there is no dumps and it just turns off and reboots.

Well, Windows doesn't actually restart (or even turn the PC itself off, as seen in the above video) - but yeah, makes sense.

I mean... I normally wouldn't exactly want a bluescreen, but it's kinda weirding me out that I'm getting only getting "black screens" and NO bluescreens, like Windows is skipping that middle-step.



...Is there a program that can deliberately (and safely!!!) cause a bluescreen on command, as in "press this button to cause a BSOD"? I'm kinda suspicious to see if that would cause a bluescreen or just another black screen instead.

HedgeToaster said:

I've made a small 70 GB partition (better to have some space just in case).
I've downloaded the MediaCreationTool22H2 ( https://www.microsoft.com/en-us/soft...load/windows10 or https://www.microsoft.com/de-de/soft...load/windows10 ), and am gonna use a USB stick for this.


Just to be clear: Of course I don't want this to mess with the already existing OS, or delete any of my files.
Will the installer give me the option to choose where to install Windows 10, and it'll leave the other partitions as they are?
Is there a specific section in the installation I should pay special attention to, or anything else I should know about? At which point do I choose the partition to install Windows to?

Yes it will let you choose the partition where to install windows without touching the other partitions.



On this image of the windows installation procedure, you click on "custom install only windows (advanced)" it is after having clicked that you will arrive on the window of the choice of the partitions where to install windows. Choose the partition you created and click "next" at the bottom right.

HedgeToaster said:

Well, Windows doesn't actually restart (or even turn the PC itself off, as seen in the above video) - but yeah, makes sense.

I mean... I normally wouldn't exactly want a bluescreen, but it's kinda weirding me out that I'm getting only getting "black screens" and NO bluescreens, like Windows is skipping that middle-step.

...Is there a program that can deliberately (and safely!!!) cause a bluescreen on command, as in "press this button to cause a BSOD"? I'm kinda suspicious to see if that would cause a bluescreen or just another black screen instead.

Trying to force a blue screen would not achieve anything.

Those images you relayed say Kernel power with a error code 41 which means the computer has shutdown unexpectedly. Whether the computer actually shutdown or not that error can happen if the computer just hard hangs which is what is happening here, it can also happen for various other reasons like overclocking which is possibly irrelevant to you but just saying its not clear cut grey and white.
https://social.technet.microsoft.com...ent-id-41.aspx

Generally if you get this error you look at what is around it because that is usually in conjunction with the problem. You have volmgr, kernel power, vbox and tablet service.

You also have a win32k warning which is to do with the windows subsystem like drawing stuff on the screen the gui and the mouse cursor and windows etc. This probably has to do with the computer black screening maybe not directly but that this is the point where the computer black screens.

Again i would suggest to run the BSOD tool found in the respective forum for BSOD's and then paste your logs in there by starting a new thread. There is one user here who is pretty knowledgeable on this stuff. I guess you wont really find out much otherwise and its a guessing game atm.
Speaking for myself i really only know the basics of this stuff as its at the pinnacle of computer science i am still learning how to read stuff like this.

go here https://www.tenforums.com/bsod-crashes-debugging/ run the tool and then upload your logs. it will be easier for people to help you that way and like i said there is one user who is knowledgeable on this stuff.

Malneb said:

go here https://www.tenforums.com/bsod-crashes-debugging/ run the tool and then upload your logs. it will be easier for people to help you that way and like i said there is one user who is knowledgeable on this stuff.

Thanks, I'll be looking into that, need to type up a post for that other sub-forum first.

Malneb said:

Generally if you get this error you look at what is around it because that is usually in conjunction with the problem.

My first thought too of course.

One thing I do want to point out though: All the Event Viewer entries you mentioned happened on restarting the PC after the crash, not "right before the crash".

I admit, I've only recently started looking at the "Windows Logs > System" section of the Event Viewer, as opposed to just "Administrative Events". However, I haven't seen any obvious entries that happened right before crashing. Most of the time, the most recent entry before a crash happened minutes prior.

The most extreme case is a crash where the Event Viewer has NOTHING preceding the crash at 13:22:





(The PC was turned off at ~4:30 in the morning, then booted up around ~13:22. Then again on 13:28.)

...And I checked the "Windows Logs" sections too, the only entries I could even find were under "Windows Logs > Security", and it was just generic phrases like "Special privileges assigned to new logon", "An account was successfully logged on", etc... which doesn't seem related, I think.



This is actually the Event Viewer log from the crash shown in the video here, while no peripherals other than keyboard and monitor were attached. (Also no internet.)



In that case, there isn't a single entry in Event Viewer between bootup and crash, except "Windows Logs > security". All other logs only recommenced after restarting the PC after the crash.

So... yeaargh.

Malneb said:

You have volmgr, kernel power, vbox and tablet service. You also have a win32k warning [...]

"volmgr" and "kernel power" are definitely referring the crash that happened earlier, however to my (admittedly untrained) eyes, "vbox" ("Oracle VM Virtual Box", perhaps?), "tablet service" (presumably my 2008 Wacom drawing tablet, or Windows's built-in drivers for handling tablets), and "win32k" (maybe the mouse?) happen a lot more frequently on bootup (always? I'd have to double-check that), this isn't usually followed by a crash. And crashes aren't limited to bootup either. So my guess would be to dismiss these as unlikely culprits. I may be wrong of course.

- - - Updated - - -

itsme1 said:

Yes it will let you choose the partition where to install windows without touching the other partitions.

Okay, thanks! :)

NTN said:

OCCT, a free CPU/GPU/PSU stress testing tool. OCCT stress tests your CPU, GPU, and motherboard, and in the process, can discover any faults with your power supply unit.
How to Fix a Windows 10 Kernel Power Error in 5 Easy Steps

@NTN, I've installed OCCT, but haven't run it yet. The test I should be running is "Test -> Power", right?



Default duration for all tests was "infinite / Test will run until cancelled". ...Yeah no.

What duration is recommended? 10 minutes? 30 minutes? Again, is OCCT safe?

I've also noticed that in Settings > Security, the default for "stop on error / critical temperature / stop of WHEA error" is "disabled".

I think I'll turn all those on, just to be safe. "Stop any running test 90°C" is the default temperature for that, that's sensible, right?



It would be quite idiotic if I damaged my system in pursuit of finding errors. ^_^;;


Is running "Test -> Power" enough, or should I run other tests too? Are there any other settings I should change from the default?

Better look at the experts.
There are others too.....maybe better.

Why not updating your old version 1909 to the newest one?
Maybe it solves a possible software issue.
Further I assume that you do not overclock your cpu?

PiKo said:

Further I assume that you do not overclock your cpu?

You are correct, I don't overclock my cpu.

PiKo said:

Why not updating your old version 1909 to the newest one?
Maybe it solves a possible software issue.

Too late, already installed a second Windows 10 copy. ^_^#

In any case, the Customer Support told me to "reinstall Windows to fix your problem", so I guess I gotta try that first anyway if I want further help from them.

Obviously I wouldn't just bulldoze my current PC, so itsme1's suggestion to use a separate partition instead was a welcome compromise I hadn't thought of. :)

Secondly, the situation here is isn't "My PC crashes are so annoying, how do I make it stop doing that", it's "This isn't so bad yet, but is it indicative of a serious hardware issue I need to get fixed before my warranty runs out?"

If it's "just" a software issue, I can relax, no need to hurry to fix this within warranty.

A fresh copy of Windows has less complicating factors to muddle things and might make it easier to narrow down the cause of the issue.

NTN said:

Better look at the experts. [...]

Hmmm. This seems to be specifically about testing the CPU for errors only.

I did have a full 6 hour run of a different program, "Hot CPU Tester 4 Pro (Lite Edition)", with no errors.




(Well okay, "no errors" is a simplification, there was a crash after 2 hours the first time I ran it, but that was because of additional added factors rather than the program alone. Reproducing that combination of factors was in fact a reproducible crash case.
But running the program for the intended 6 hours with a normal workload worked fine, and reported finding no errors.)


I ran the following OCCT tests, for 15 minutes each:



No errors or crashes during those 15 minutes. Not sure if that was long enough, but also am a bit reluctant cause I don't know if it could physically damage the hardware.

Not sure if I should re-run the "Power test" for longer...?

Okay, several updates.

First of all, I realize something really stupid happened. I had never installed the msi (motherboard) or AMD (GPU / CPU) drivers when I received the PC.

So... yeah that's probably related to the problem in some capacity.

This was my first time ordering a custom PC online, selecting components on a website and having the assembled PC shipped, instead of physically picking up a pre-made system at a store; I guess I just assumed "oh I bought a new PC with Windows 10 on it, everything to make it run must already be installed."


Did that fix it? Ehhh... kinda hard to say.

Not initially, no - but when I checked "Control Panel\All Control Panel Items\Security and Maintenance\Reliability Monitor", I saw:

"Finish installion device software - One or more devices connected to your PC needs additional software to work properly."

Clicked on that, and... my "reproducable crash tests" no longer crash the system, so that's a change.

There's a new Power Plan now, "AMD Ryzen Balanced", instead of the regular "Balanced (Recommended)" that 100% definitely wasn't there before according to my screenshots. So if the issue was related to the power supply, the prior lack of this might explain it.


Was already thinking "horray! That fixed it!"

But the system (still my regular old Windows, NOT the fresh install!) still randomly crashed afterwards, three times in the same day even, as if to prove "lol nope". (Twice on bootup after turning the PC off for a few minutes, once randomly in the evening during normal load. All crashes on the same day though, and none since.)

___________________________________________________________________________________________________


Okay, but what about that fresh Windows 10 installation on a separate partition?

The jury is still out on that one.

It actually managed to crash, once, same old "no video signal" / "no Event Viewer entry", but that was before I had even managed to install the msi & AMD drivers on here too.

Hasn't crashed since I installed those (and updated Windows), so that's a hopeful sign.


Also re-seated the RAM finally. (AFTER the last random crashes had happened, that is.)

Didn't manage to eject the graphics card to re-seat that, the way it was built makes it extremely dangerous to try pushing on the tiny little sliver of an "eject"-button on the motherboard sticking out from under the graphics card, even if all I'm using is my finger and a piece of strong cardboard – the risk of slipping off and breaking something off from the motherboard is too great. I sorta wiggled the left side of the graphics card a few millimeters out of its seating by accident (obviously after unscrewing the screws and unplugging the PCIe cable), and then clicked that back down, that's gotta be good enough as a "partial re-seating".

___________________________________________________________________________________________________

Does that mean the problem is fixed? I don't know yet. No crashes since then, or "not yet", on their own they do happen randomly and very sporadically. Don't know any other "reproducible crash test" I could do to make sure. Probably won't know by the time my warranty expires, which is, errr, on... January 22nd. ^_^;;


But at this point I'm guessing this looks more like a software problem rather than a hardware one, doesn't it? In which case... this would be fine.

___________________________________________________________________________________________________
- - - Updated - - -

Okay, it's a hardware problem.

The freshly installed Windows just crashed within 5 minutes of booting up.

Also, one of my two crash tests turned out to work on the freshly installed Windows after all.


The chronology today was this, all on the freshly installed Windows:

1) Run crash stress test: 2 hours of CPU stress test (physical cores at 100%, virtual cores left alone), otherwise leave PC alone for 2 hours, then suddenly start a bunch of programs and other stress tests all at once to see if it causes a crash. Last time I did that, on my main Windows, it no longer crashed. Turns out now it does on the new Windows within less than 5 minutes.)

2) Restarted PC, checked some stuff, then turned it off again.

3) Left PC off for an hour. Just turned it on and wasn't doing anything in particular... and then it crashed within the first 5 minutes.

4) Restarted the PC afterwards, has been up and running for 30 minutes now

So yeah. -_-

Is BIOS up to date?

Does it crash or is more like it is freezing (the clock on taskbar also freeze..)?
If so it could be the GPU...perhaps.