Urgent! PC crash to randomly coloured screens (no BSOD), requires hard reboot

Xav101

Prominent
Jul 1, 2017
27
0
530
For about a week my PC's been having this issue. While playing games both of my monitors will go black for about half a second before they come back showing a solid colour (different on each and seems random, I've had pink, green, red, blue, white). Sound continues to play although it's not what it should be (e.g. can't hear discord, can't hear shooting when I should be). I've read many threads about this issue but none have been able to help me.

This problem has occurred while playing games from both my SSD (Rainbow Six: Siege, CS:GO, Dead By Daylight) and hard drive (Assassin's Creed Origins). Although it runs fine outside of gaming. Siege seems to be pretty consistent at triggering the crash relatively quickly so I've been using that to test after making changes.

The fact that this occurs on both monitors makes me doubt that this is a monitor/monitor cable issue.

Event viewer shows nothing other than the occasional "The system has rebooted without cleanly shutting down first." due to having to hard restart. It has told me that "display driver nvlddmkm stopped responding and has successfully recovered" but I seem to have fixed that.
I used to get system notifications that say "Display driver has stopped responding and is now using basic drivers" (sorry, I can't quite remember it but that was close). But haven't gotten this notification for a while.


TL;DR
Monitors turn to a solid (random colour) while gaming. Controls and PC unresponsive. Sound sometimes unaffected, other times it's the wrong sounds. Requires hard reboot using power button.

What I've Done
- Used monitoring software (MSI Afterburner and HW monitor) still crashes while GPU and CPU temps are under 70 degrees
- Also tested with Afterburner and HW monitor uninstalled
- Removed all GPU overclock
- Tested base CPU clock
- Re seated GPU, in same slot and in lower slot (still crashed)
- Re-applied thermal paste on GPU and CPU
- Ran Memtest 86, no errors
- Left Heaven Benchmark running for about 4 hours without problems
- Disconnected disk drive (read that it helped someone before)
- Unplugged and plugged monitors back in after crash and before shut down
- Reinstalled Nvidia drivers multiple times (just through Geforce Experience, complete uninstall in
device manager then use Geforce Experience, complete uninstall in device manager then use
Nvidia website, use DDU then use Nvidia website).
- Reinstalled windows (keep apps & files, then again using removing everything)
- Run Malwarebytes scans, registry cleaner, avast scans and chkdsk
- Verified game files of Rainbow Six: Siege

I do have a gt 610 that I could use in place of my current GPU if needed but I haven't been able to withstand going from a GTX 970 to it and playing more than 1 or 2 games of Rainbow Six: Siege on it.

At this stage I'm unable to use my PC's full potential and am beginning to run short of ideas. Any help would be greatly appreciated :)
- Xavier


System Specs
OS > Windows 10 Pro version 10.0.17763 build 17763
CPU > i5 3570K @ 4.2 ghz and stock (3.4 ghz)
Motherboard > ASRock Z77 Extreme4-M
GPU > Gainward GeForce GTX 970 Phantom
RAM > 16 GB (2 x 8) DDR3 Corsair Vengeance Ram (not sure what type)
PSU > Corsair CMPSU-620HX, 620W
SSD > Kingston A400 240GB
HDD > WD Scorpio Blue 500 GB

Main Monitor > benq xl2411t @144hz
Second Monitor > AOC TFT24W80PSA
 
Solution
I dont think that this is a Monitor Issue.
I would rather say that your GPU is getting old and starts to die.
Had same Issues with some GPUs in the past were everything was fine in desktop but i´ve you load up a game or a Stresstest.... Hell ive had some crazy flashy images :D All the way to instant black screen and automatic reboot. Then everything was fine. 5 Hour Desktop and coping files was okay but man if you start a 3D software they freaked out...

Since you have a 3570k you can take out the GPU and "game" a little bit via the IGPU or the GT 610.
If you do not experience the same problems like these drivers issues etc then its pretty much clear.

You do not have warranty left on that card or?
I dont think that this is a Monitor Issue.
I would rather say that your GPU is getting old and starts to die.
Had same Issues with some GPUs in the past were everything was fine in desktop but i´ve you load up a game or a Stresstest.... Hell ive had some crazy flashy images :D All the way to instant black screen and automatic reboot. Then everything was fine. 5 Hour Desktop and coping files was okay but man if you start a 3D software they freaked out...

Since you have a 3570k you can take out the GPU and "game" a little bit via the IGPU or the GT 610.
If you do not experience the same problems like these drivers issues etc then its pretty much clear.

You do not have warranty left on that card or?
 
Solution

Xav101

Prominent
Jul 1, 2017
27
0
530


I don't have a warranty on the card as I bought the tower second hand about a year ago. And I'll play maybe 10 matches of siege on the 610 to test that and report back.

If it turns out that it's the card/drivers is it possible to diagnose which then fix it? I forgot to mention before that I've used all 3 recent GTX 970 drivers.
 
You can run diagnose tools and yes it maybe help.

The problem i have in my mind is, that the gpu itself does have (some) cold soldering points.
This (often) happens between the DIE and the Chip the DIE sits on (or between the Chip and the actually PCB of the GPU). The soldering points can lose contact and this can lead to articafts, bluescreens, failing drivers and and and...

There is no long term fix for this (other then a complete reball of the GPU). Some people put their GPU in the oven at ~150 Degree C for 30-45 minutes. This CAN bring the GPU back to life for a few months (most people were talking about ~3 months) before the problems come back.
 

Xav101

Prominent
Jul 1, 2017
27
0
530
Ok, I’ve finished the testing; 10 matches of siege (3 continuous matches yesterday and 7 today. As well as a few runs of the in-game benchmark) in the gt 610 without any crashes which makes me almost sure that it’s an issue with the card.

I’ve seen a few videos on your baking method, but I’ve only got the one oven (for food) and seeing as it’s only a (very) temporary fix that’s not even definite, I might end up getting a secondhand 1070 for Christmas for 350-400 aud, unless there IS a fix or a more reasonable card, of course.
 


1070 is a great card. 1060 would be about the same performance as a 970 :)

You can try to get 980/ti seconds hand they are also great.