[SOLVED] Really Weird System Behavior

[pindiespace]

I have an older gigabyte Z170x Gaming 6 system, SSD only, lots of RAM, H100i water-cooled. It occasionally crashes in an extremely strange way:

1. OK for weeks or even 3-4 months
2. Sudden crash, normal reboot
3. Crashes become more frequent, always reboots OK
4. Crashes are 1/2 hour, 5 minutes, apart. Sometimes crash asks "do you want to proceed to Windows 10"
5. Final crash - I start it, it crashes 3-10 seconds later, reboots itself....
6. OK for weeks to months

I can't think of anything that would behave this way. It is as if some sort of timing happens between a crash event. and the interval shortens. Somehow, when the crashes are a few seconds apart, it resets to a long interval.

The one other problem - the intel CPU is listed as "unsupported" by Windows 11, so I have never been able to upgrade to Win11. Howver, this strange crash behavior started long before that. I've wondered if there is some strange pump problem leading to temperature spike - but crashes are uncorrelated with the temperature Corsair app reports for the system.

This is a job for a real guru of hardware/software

I've tried taking everything apart and cleaning it, no dice, even brushing out the power supply .

It seems like a weird timing issue. I could replace the board,

 

[Darkbreeze]

What is your CPU model? FULL hardware specifications? Motherboard BIOS version?

We can go from there including checking whether you've actually manually kept all drivers up to date including chipset (.inf), audio chipset, storage controller, etc. Need to know all hardware specs though, preferably with EXACT model numbers, before we can even begin to determine anything.

 

[pindiespace]

CPU: Intel Core i7 6700K
Mainboard: Gigabyte x170x Gaming 6 LGA1151 Bios F3
Video: NVIDIA GeForce GTX 1070
RAM: Corsair CMK32GX4M2B3200C16 - 64GB
Liquid Cooler: H110i v2
Storage: Samsung SSD870 QVO 2 TB

 

[Richj444]

CPU: Intel Core i7 6700K
Mainboard: Gigabyte x170x Gaming 6 LGA1151 Bios F3
Video: NVIDIA GeForce GTX 1070
RAM: Corsair CMK32GX4M2B3200C16 - 64GB
Liquid Cooler: H110i v2
Storage: Samsung SSD870 QVO 2 TB

What model is your power supply? And how old is it?

 

[Darkbreeze]

Yes, exact model of power supply is a fundamental necessity. Also, in the future, do NOT "brush out" the power supply. Simply blow it out with compressed air. Opening up your power supply and playing around inside it is a good way to end up dead or seriously injured. And yes, even when it's not plugged into anything, it is still extremely dangerous if you don't know what you are doing.

 

[pindiespace]

Yes, exact model of power supply is a fundamental necessity. Also, in the future, do NOT "brush out" the power supply. Simply blow it out with compressed air. Opening up your power supply and playing around inside it is a good way to end up dead or seriously injured. And yes, even when it's not plugged into anything, it is still extremely dangerous if you don't know what you are doing.

Sorry I meant used air to clean the power supply not physical brush.

I didn't suspect the power supply because of the strange timing of crashes - far apart, shorter and shorter, a few seconds, than back to weeks or months.

 

[pindiespace]

What model is your power supply? And how old is it?

It's several years old, 1200 watt. Have to do some digging to determine exact model. I didn't consider since the weird timing pattern seems inconsistent with a malfunctioning power supply.

 

[Darkbreeze]

EVERY problem, can be the power supply, because EVERY problem could be a piece of hardware that RELIES on the power supply to function correctly.

 

[pindiespace]

EVERY problem, can be the power supply, because EVERY problem could be a piece of hardware that RELIES on the power supply to function correctly.

OK, so it makes sense to start with the power supply. So, I need to figure out a power supply that

(1) compatible, has all the plugs I need for the Z170x Gaming 6 system (many don't).

(2) Old supply was 1200w, probably overkill, 1000w or a little less should be fine. I used to run a Rift through it, but now I have a quest, don't need the host computer anymore, so the graphic card barely breaks a sweat on large Figma files.

(2) PSU either has very long power cables, or the power supply itself is deeper that a regular ATX. The old power supply is 10 inches deep, and the cables barely reached. I checked with a standard -depth ATX, and the cables with the z170x didn't reach the PSU. Case is just a full-sized coolermaster tower.

Despite the psu importance, you rarely see compatible PSUs listed in "builds"

 

[Darkbreeze]

All you need for a system with a GTX 1070 is a good 550-650w unit. Anything beyond that is just gravy and is unnecessary capacity for the needs of the system unless you have plans to upgrade to a newer high end GPU in the near future since those cards are power hungry, much more so than cards from your card's generation. Key being "good" 550-650w unit. If you buy a mediocre model you'd want it to be more like 750w but I wouldn't recommend ever buying a poor or mediocre quality PSU for anything more than a basic internet browsing machine with integrated graphics. And probably not even then unless there was literally no choice. It's the most important component in the system after all.

So, what IS the model of your current power supply? This is relatively important and if you are unsure the model will be listed on the label on the side of the PSU. All power supplies have one, it is required by law in pretty much every country.

 

[pindiespace]

Existing power supply is OCZ ZX Series 1250 watt. Was overkill at the time (2012).

 

[Bob.B]

Existing power supply is OCZ ZX Series 1250 watt. Was overkill at the time (2012).

At 10 yrs old that's double the warranty period.

Perhaps it's time to think about a replacement.

 

[Darkbreeze]

Yeah, double the years of service as the original warranty is a definite potential recipe for disaster. It's was a very good unit 10 years ago, but six years ago somebody should have begun looking for a replacement especially if they were going to run rather expensive hardware with it.

I'd definitely replace that PSU.

As to temperatures, I would suggest that you not use the Corsair app, and instead use HWinfo which will allow you to see all relevant CPU and graphics card thermal sensor readings. Download it, install it, choose "Sensors only" and uncheck "Summary" and then run it while doing something demanding. Stress CPU then stress GPU to see what thermals do and what is reported in HWinfo.

 

[pindiespace]

OK, this looks like a place to start! I'll look at the PSU model recommendations above, thanks!

 

[Darkbreeze]

👍

 

[pindiespace]

Turns out the problem was with the motherboard, after all. Replaced, and everything perfect, even with old PSU.