[SOLVED] Random shutdowns led to corruption of Windows - secondary OS works

Status
Not open for further replies.
Aug 31, 2021
4
0
10
This situation has evolved quicker than I had anticipated. I will break it down in detail below, but here is a TL;DR with PC specs and some notes.

=======
TL;DR
Primary problem: random crashes; PC shuts down, as if I pulled the plug; Cannot boot for 15-20s; Everything is fine afterwards. Number of such crashes: 6
Secondary problem: GPU-bound crash; black screen; "No signal"; GPU fans at 100%; OS is still running in the background; audio is being received; Only solvable by hard shut down (physical button). Number of such crashes: 1
Tertiary problem: Cannot boot into Windows; no Windows logo or loading screen; A line of seemingly "dead" pixels; BIOS screen is fine; Booting into secondary OS (Ubuntu) doesn't reproduce the issue. Comparison videos for reference: Booting to Windows and Ubuntu 20.04 respectively

What I have already tried:
  • Checked the PCI-e cable = normal
  • Checked the power cable = normal
  • Checked for PSU overheating issues = none
  • Checked to see if PSU fan works = normal
  • Checked for CPU and GPU overheating issues = none (60-70°C)
  • Checked CPU and GPU fans = no coil whine, work as intended (except for the case in the secondary problem mentioned in TL;DR)
  • Checked for malware = none
PC Specs:
BIOS: UEFI
Motherboard (New): MSI B450 Tomahawk Max
CPU (New): Ryzen R5 3600, stock cooler
RAM (New): 2x8GB Crucial Ballistix DDR4 3600 MHz
GPU (used): Asus Strix GTX 970 4 GB GDDR5
PSU (New): Silverstone 650W 80+ Gold (SST-ET650-G)
SSD (New): Samsung 860 QVO 1TB 2.5"
HDD (3 yrs old, bought new): Toshiba HDWD110 1TB 3.5"
OS: Dual Boot (Windows 10 + Ubuntu 20.04) - both on SSD

Current status: I am bound to Ubuntu for the moment. GPU works as intended. Cannot boot into Windows. EDIT (12:35): Cannot turn on the PC. Before this, audio was crackling/had a bitcrush-type effect. Edit (12:43): It turned on again, sound is back to normal. I`m going to copy my important files to an external drive just in case.

Priority: My main priority was to solve the crashing problem, but now I want to fix Windows to boot again. What is the recommended course of action? I hope Ubuntu doesn't get corrupted as well...


Notes:
  • Even though my PSU is rated at 80+ Gold, it brings continuous power output only up to 40°C, which is lower than other PSUs in the market at this rating, from my research.
  • Excluding the GPU, my PC is relatively new (about 6 months)
  • Between the first and second crash (previous month vs. Saturday), there were two power outages back-to-back, which made my PC crash two additional times in the same fashion, but I excluded this, since it is connected to the power grid.
  • Edit (13:43): Disconnecting an external hard drive causes a crash in the same fashion as the one in the primary problem.
Hypothesis: Since the GPU seems to be working completely fine, as well as the CPU and considering the fact that the PSU delivers continuous power output only up to 40°C, I have strong suspicions that IT might be the root of all of my problems. Even though I have two more PSUs at hand, they bring only up to around 300 or 400W of power, which is not enough for my system (they are also those janky grey power supplies), so I cannot test that out (I would have to buy a new PSU).
=======
Detailed description:

Previous month: My PC seemingly randomly crashed for the first time. I think I heard a click of the power supply (which is normal, as it also happens, when I shut down my PC, but not so much it crashing). There was nothing crazy, the screen just went black and everything came to a halt. I wasn't able to boot it up again for maybe 15-20s, but afterwards, it was fine, as if nothing had happened. Because of the timeout, I thought the PC was trying to cool itself, so I checked my CPU and GPU temps with a Blender classroom benchmark (GPU usage was at 100%, CPU at around 20%), but they were fine (60-70°C). From Googling, I was suspicious of my PSU, which is the weakest link in my PC, right after my GPU (because I bought it used - 4 years of previous uptime). Nevertheless, I continued using my PC with caution.

Saturday 28. 08. 2021 and Sunday 29. 08.: Then, the previous Saturday and Sunday the same thing happened. I saw the trend. The crashes were happening more frequently.

Monday 30. 08. 2021: Then, yesterday happened. I had it crash after 30mins of a trivial PC task (playing back a movie) in the morning and once in the afternoon (around 1 p.m.), while doing practically nothing. In the evening, I booted up Linux to try it out for a gaming spin with my friends and it too suffered the same fate. I thought it was overheating, so I opened my window and went back to gaming for another 30 mins or so (very foolish). 5 minutes after I closed my window, a new symptom showed up and a new crash. My screen went black, but I could still hear my friends over Discord. Then my GPU fans started howling like crazy (they were running at 100%, then slowed down to maybe 80% for half a second then back to 100% and so on in an aggressive sinusoidal fashion). I turned off my PC with the physical power off button and it shut down. I didn't touch the PC for the rest of that night (I thought: "Cool, atleast I didn't yank the power cable from the wall, so nothing could be wrong right?").

Today: I had plans to test my PSU and GPU to see, which one is causing this havoc, but they were thrown out the window, when I couldn't boot into Windows 10. I see the BIOS screen, but when the Windows logo should show up, there is just this line of spaced-out green pixels and some below it. You can see for yourself in the TL;DR section. I can boot into Ubuntu (dual boot on the same SSD) just fine though, so this can only be a sign of a corrupted boot loader or the Windows partition or something of that sort.
 
Last edited:
Solution
Are both installs on the ssd?

All the crashes could have corrupted windows install.

Win 10 by default uses a mode called fast startup. With it on, a windows 10 PC is never off except during the restart operation (in power options). Its in a hybrid hibernate mode where some files are saved in ram and some are saved on hdd. When you shut pc down it actually goes to sleep.

What has that got to do with your case? Well, windows when it crashes doesn't shutdown properly (obvious) and this can corrupt files needed to boot the PC.

So you might need to clean install win 10 to get it back.

IF problem occurs in Linux and Windows, its not software

could be GPU itself how old is it? My 980 died last year 🙁
Are both installs on the ssd?

All the crashes could have corrupted windows install.

Win 10 by default uses a mode called fast startup. With it on, a windows 10 PC is never off except during the restart operation (in power options). Its in a hybrid hibernate mode where some files are saved in ram and some are saved on hdd. When you shut pc down it actually goes to sleep.

What has that got to do with your case? Well, windows when it crashes doesn't shutdown properly (obvious) and this can corrupt files needed to boot the PC.

So you might need to clean install win 10 to get it back.

IF problem occurs in Linux and Windows, its not software

could be GPU itself how old is it? My 980 died last year 🙁
 
Solution
Are both installs on the ssd?

All the crashes could have corrupted windows install.

Win 10 by default uses a mode called fast startup. With it on, a windows 10 PC is never off except during the restart operation (in power options). Its in a hybrid hibernate mode where some files are saved in ram and some are saved on hdd. When you shut pc down it actually goes to sleep.

What has that got to do with your case? Well, windows when it crashes doesn't shutdown properly (obvious) and this can corrupt files needed to boot the PC.

So you might need to clean install win 10 to get it back.

IF problem occurs in Linux and Windows, its not software

could be GPU itself how old is it? My 980 died last year 🙁

The crashes definitely corrupted Windows. I did have hibernation on, so that probably contributed to it (maybe had I booted it up immediately after like I used to, there would have been programs still in the memory and I would've avoided this).

The crash under category "primary problem" happened once in Linux, but I didn't use it for gaming since yesterday, as to minimize the risk of another crash. Edit: I found a system log on Linux, which stated that at the time of this crash, the GPU "had fallen off the bus".

The GPU is supposedly 4-5 years old, but it seems to be working fine at idle in Linux. I am omw to pickup a brand new PSU atm. Once I install everything correctly, I'll try a stress test and a gaming test to see if the problem still persists.

I'll hold off reinstalling Windows, because I have some important files I need to backup.

Edit: Yes, both installs are on the SSD.
 
Last edited:
I didn't know Silverstone made their own PSU. I used their cases for 12 years and would have liked one of their PSU in the case.

That one is an entry lvl model - https://www.tomshardware.com/news/silverstone-releases-three-essential-psus,35605.html

Even though my PSU is rated at 80+ Gold, it brings continuous power output only up to 40°C, which is lower than other PSUs in the market at this rating, from my research.
do you mean operation temps as some pretty good CPU have same range, for example - https://seasonic.com/focus-gx?__cf_...RDPozADWp60-1630430505-0-gqNtZGzNAdCjcnBszQiR

My 980 was 5 years old.
 
Last edited:

I bought it because it was cheap (like 70€ where I live), which is in hindsight the worst decision I could've made.

do you mean operation temps as some pretty good CPU have same range, for example - https://seasonic.com/focus-gx?__cf_...RDPozADWp60-1630430505-0-gqNtZGzNAdCjcnBszQiR

Perhaps, but I don't reckon they work as efficiently than for example ROG Strix RM750 (0-50°C operating temps), which I have bought and installed now. The previous PSU might have been faulty at some level, since at startup, I would hear the GPU fans ramp up before ramping down (loading drivers I supposed), but now with the RM750, it doesn't do that anymore. I did some stress tests again on Linux and no crashes occurred. I just need to recreate what happened before my second crash to be absolutely sure.

Just in advance, since I probably am going to have to reinstall Windows: Can I reinstall it without wiping everything off my SSD (hinting at keeping Linux as a secondary OS). Is there an option in the bootable Windows Media Creation Tool that allows me to install Windows on a particular partition?
 
I ended up reinstalling Windows and backing up most of my files (although annoying, it was the only option). I replaced my old PSU with an ROG Strix RM750 and had no crashes as of yet. Thank you all for your kind help!
 
Status
Not open for further replies.