Question Constant crashing after gaming/tabbing out

[Nurki]

Hi, hoping this is the right area of the forums, if not please let me know and I'll make a post in a more fitting area.

I have an issue with constant crashing that I can not figure out on my own. First, my specs:
Nvidia GeForce GTX 1070 8 Gbps
Intel Core i7-7700K Kaby Lake CPU, 4.2GHz
Corsair Vengeance RAM 3000MHz, 32GB, XMP
ASUS ROG Strix Z270H Gaming MOBO
Samsung 970 EVO 1TB m.2 disk
EVGA SuperNOVA 1000w P6 Platinum PSU
Windows 10 Pro

My pc crashes like clockwork. If I tab out once or twice after playing games for a while my PC will crash. It has NEVER crashed whilst I was in game and had the game open, only if I tab out. If I end my gaming session and close the game I have anything from a couple seconds to 1 minute before it crashes. Screen goes black and I get "no signal" up on my monitor, but the fans in the case keep running and my MOBO has the on-light turned on. If I have a video or music running on youtube it will sometimes keep playing the sound normally for up to 10 seconds, and sometimes the sound just starts lagging. If I leave it for a minute it will restart itself, and it's clearly been a dirty shutdown as I have to restore my web browser etc and event viewer telling me shutdown was unexpected. There are 0 entries in dump files apart from saying the shutdown was unexpected around the time of the crash.

When it crashes I have to wait a few minutes to restart or it will just instantly crash again unless I can open a game in time. I have had a couple crashes without having a game open, those times I had a youtube video running and was browsing/googling in addition to that, so my pc is at that point basically idle.
So strange how it's MORE stable whilst in a game than it is when I am on my desktop/browser.

Things already tried after the first 40 something crashes:
Gutted pc, fully cleaned and re-seated everything apart from the CPU
Fully cleaned cpu fan and applied new high quality thermal paste
Fresh windows install and thus also fresh game installs
Removed all and any bloatware
Tried repairing/file integrity check on everything and its grandma
Removed hardware acceleration on everything that might use it
Lowered graphics quality in games, increased graphics quality in games, v-sync on and off
Updated drivers, BIOS and Chipset on everything
Tried a soft overclocking through the motherboard presets
Ran assorted benchmark software to make sure CPU, GPU, M2 disk (Samsung Magician) is working without fault, all came back in good shape with no errors.
I have also tried a few different software to read dump files, reliability logs etc, no trace of any errors.

New as of TODAY: Brand new PSU and wiring

And after all of this poking and prodding ... Still crashing.

No temperatures are going high even under max benchmark load. Peaked at 71 celcius on CPU and 68 celcius on GPU, everything else below 50 celcius.
XTU2 benchmark sets my CPU at "1881 marks", "4,5ghz max CPU frequency", "71 celcius max temp". CPU memory stress test: passed
RAM test: passed, no errors
No LED warning lights show on the MOBO during or after a crash.


Am I wrong thinking this is a hardware issue, GPU or CPU? I can't find anyone else with this exact issue after hours of googling and reading forums.

 

[Ralston18]

Look in Reliability History and Event Viewer.

Either one or both may be capturing some error codes, warnings, or even informational events just before or at the time of the crashes.

Start with Reliability History. Much more user friendly and the timeline format may reveal some pattern involved.

Pay attention to the error codes. The codes may or may not be helpful.

 

[Nurki]

There are no error codes in the time of a crash. I have checked all the logs and in addition also used software like WhoCrashed etc to make sure I haven't missed anything glaringly obvious. The literal only error in there was from a month or so ago when the CPU got too warm after the fans ramped down(boiling hot inside, I had been gaming for hours and the fans weren't efficient enough) so the system shut down when the CPU was at 95 celcius to prevent hardware damage. At that point I had already been crashing for a couple months. I did a full re-build of the pc after this and the system has not passed the stated temps in the original post. And prior to that the highest temp was 89 celcius on the CPU after hours of load. I monitor general temps with HWMonitor.

There are errors upon startup after a crash saying the shutdown was unexpected, and something about 'windows hello for business' isn't able to launch/update. Nothing at all before crashing. I would assume if it was a driver or something it would leave some sort of trace.

I really hoped the new PSU would sort it, that it might just have been a power surge from the GPU or something when closing a game/alt-tabbing and changing modes/fan speeds that my old Corsair 550w Gold PSU couldn't accommodate.

I am off on holiday tomorrow morning, if there are any suggestions I am happy to give it a go when I get back next week. I have a friend coming on the 15th of July with his old pc to test it by switching part by part to see if it's a hardware issue. Just ran out of other options to try at this point.

 

[Nurki]

Welp, after testing, scouring event logs and dump files- and getting absolutely nowhere- I got my friend to bring his old pc over to change part by part. My gut feeling was the GPU dying in the strangest of ways after 6 years of service, so started off with that.

I am on day 3 of irritating the heck out of my pc and haven't crashed once with the borrowed 1080-TI GPU.
Not even whilst having multiple games running, tabbing in and out and then closing all of them suddenly after a lot of load- something that would 100% crash my pc within a minute with the old card in. My screen that I thought was damaged and burnt in on the top 1/3 of the screen is now looking brand new. And with it being a much better card I am finally getting 0 screen tearing and no dropped frames as an added bonus.

Now to decide which card I should go for to upgrade my old and apparently also broken 1070.