[SOLVED] Unstable PC

[OuaTiSR]

Hi everyone,
I am looking for your opinion and expertise.

Recently my PC has become very unstable.
I have been building PCs for over two decades and always had spare parts laying around to test if a problem came up in the past.
Few years ago I got fed up with all that, sold all spare parts and that left me only with my current 4 years old setup
i7700K - on stock (has been overclocked in the past but only for a brief period of time)
MSI Z270 Carbon Pro
16GB (2x 8GB) HyperX 3000 MHz
MSI RX480 Gaming X 8GB
Corsair RM650x

What has happened:
the PC became a bit unstable, which seemed more like a process than a single event happening (I suspected Win10 update at the time, but it wasn't that)
problems with UE4 games (all of them crashed at startup) - legit games, no cracks
problems with audio (DPC latency) - audio distortion, screeching noises when playing videos online or in some games
occasional browser crashes (Chrome, Firefox), Edge crashing instantly after opening a webpage
interesting: PaleMoon browser experienced no crashes, no audio problems, was working stable and good
PC sometimes did not boot up on the 1st try - boot screen, then reset, then booted up like nothing happened, eventvwr showed no errors
I suspected GPU failure or problems with drivers so I uninstalled newest Radeon drivers and tried to install old drivers from 2020, after reboot
Windows was unable to start, unable to boot in safe mode, unable to repair, nothing...

What I did:
I opened the case, cleaned the whole PC, removed CPU cooler, cleaned, put new paste, reseated it, checked all connections, checked for damaged capacitors..
cleared CMOS..
then had psu, cpu and ram, without any hdd, without dedicated GPU - running on intel integrated, booted memtest86 from usb (8 passes) with no errors..
then installed 1TB m.2 ssd and tried to install Linux Mint from usb - failed during installation , tried to install Linux Mint from DVD - failed during installation
installed clean Windows 10 newest ISO from usb - no problem..
ran AIDA64 stability test - failed within seconds
ran P95 - 2 workers fail within seconds, others are stable for hours..
ran Intel Processor Diagnostic Tool - PASSED
switch memory sticks and tried them separately one by one - result is the same
browsers are crashing almost instantly - chrome STATUS_ACCESS_VIOLATION
the PC does BSOD, but only during boot, it restarts and works "normally"...
the PC never crash...only browsers.. BUT PaleMoon browser is stable, with 20+ tabs open, never crashes...all the others do (opera, firefox, edge, chrome)
tried the same on another HDD to exclude hdd error - results are the same...tested hdds for errors - result is ok..
tried upping the RAM voltage from default 1.20 to 1.35 without any effect..

based on the above - I have concluded that it is the MB failing - RAM tests are ok, CPU diagnostic tool passed...no other devices are connected..
PSU has no visible damaged capacitors, according to Aida64 and other tools the provided voltage seems ok..

Can anyone please recommend me any other tests to help me pinpoint the exact problem or a failing component (without using spare parts)?
What would you do?

 

[Ralston18]

Look in Reliability History and Event Viewer for error codes, warnings, and even informational events that correspond with the times of crashing and blue screening.

Numerous and varying errors of any sort are a symptom of power problems.

A four year old PSU ( Corsair RM650x - correct?) may not necessarily show signs of damage but could be in the early stages of failing due to designed in EOL (End of Life).

Especially if having been heavily used for gaming.....

Do you have a multi-meter and know how to use it? Or have a family member or friend who does?

https://www.lifewire.com/how-to-manually-test-a-power-supply-with-a-multimeter-2626158

Not a full test as the PSU is not under load. However any voltages that are out of tolerance make the PSU suspect.

 

[carocuore]

then installed 1TB m.2 ssd

is that new or 4 years old as the rest of the PC?
if it's a samsung or has samsung chips on it it's probably dying, it would explain why some games crash (can't load the required files or those are corrupted)
if the Linux disc failed to load any files it could also be because of a critical I/O error that causes kernel panic
could also be a mobo issue though

ran AIDA64 stability test - failed within seconds

what failed exactly? it usually shows an error like CPU overheat, unstable ALU, etc

On a side note I have a computer with a similar issue, trying to load a browser instantly causes a BSOD or a restart, but it's 20 years old and has a 1MB ISA graphics card on it, everything works except for browsers.
I couldn't solve it so I just concluded it's senile after so many years of service at the police station ^^

 

[OuaTiSR]

Thanks for the replies.

A four year old PSU ( Corsair RM650x - correct?) may not necessarily show signs of damage but could be in the early stages of failing due to designed in EOL (End of Life).

I was able to get my hands on another psu, switched it and the problem persists...

is that new or 4 years old as the rest of the PC?

the m.2 is a couple of months old and I tried the system with another hdd so the problem lies elsewhere..

what failed exactly? it usually shows an error like CPU overheat, unstable ALU, etc

Hardware failure detected! Test stopped. It does that with AVX turned on. Without it it runs stable (tried for 30 minutes or so).

I also have been experiencing errors when copying files - that is new, only since new install..
for me that means MB is erratic...
I am currently searching for a secondhand MB and will try with that..

but it's 20 years old and has a 1MB ISA graphics card on it, everything works except for browsers.

I would think that browsers crash because of low memory. You can try some lightweight browser - maybe Lynx?

 

[OuaTiSR]

Update:

So I bought used MB and tried it.
And guess what - the issue persists (however I cannot guarantee that the MB is not faulty, but I doubt it).
The only thing left to do is to use completely different RAM (I will do that tomorrow) and then the only constant used in all variations will have been the CPU.

 

[OuaTiSR]

RAM changed,
and the culprit is the CPU.
I have never seen such behaviour of a cpu, it was either faulty - not able to boot, physically damaged, bent pins, burnt etc. OR it worked as it should.
so that's it.