Various BSOD, need help please.

[Quattro]

Hi all,

Not entirely sure how active this forum still is after all these years but I need some help.
I've had my 8700k/Asus HeroX system for 2 years and never had any issues but in the last week or so I've been getting constant BSODs, kernal error, memory management, iqrl error, among others.

Thought initially it was a dodgy driver so reinstalled all the drivers on my system and that didn't make any difference so tried a fresh install of Windows10, still no cure. Checked device manager and found that "hid compliant headset not working" was being reported so uninstalled the driver and kept my usb headset unplugged, still get BSOD.

So with the memory management BSOD I thought maybe it's one of my RAM sticks are dying (G.SKILL TridentZ RGB Series 32GB (2 x 16GB) 288-Pin DDR4 SDRAM DDR4 3200 (PC4 25600) Desktop Memory Model F4-3200C14D-32GTZR) so I tested them individually in Memtest86 for four passes, no errors. Tested them together in slots 2 and 4, no errors. Tests done with and without XMP.

The PSU is not too old, maybe been using it for 10-12 months. Corsair HX1000i.
Graphics card is a 1080Ti (It has been overclocked since day 1, no heat issues, usually sits around 70-75c under maximum load) that I've had since launch so it's getting on a bit but I don't think this would cause the BSOD that I am getting?

Windows10 is installed on a Samsung 970Pro NVME m.2 drive that seems to be working correctly.

The BSODS seem to occur at random times, sometimes not for 30 minutes and sometimes I can't even download 200mb drivers without it crashing three times.
Doesn't seem to be heat related, 8700k under load is around 60-70c (Kraken X62), the overclock I've had it one for the last two years was 5.0ghz but with only 1.28vcore, I always left the memory timings alone and just used XMP defaults.

I'm thinking that possibly I've got a faulty motherboard, but wouldn't memtest86 show that?
Is there anything else you can think of?

My keyboard and mouse is a USB Coolermaster one, I forget the name, peripherals I used are Logitech/Saitek X56 Hotas, TrackIR5 sensor and software, HyperXCloud USB headset with mic.

Nothing is new, it was all working perfectly fine up until about a week ago.
Any advice or help would be much appreciated.

 

[trents]

Is the CPU overclocked? If so, try lowering the overclock a tad or raising the vcore.

With 32 gb of RAM you may need to increase the voltage to the IMC a tad. You say you tested each stick individually with memtest86 but did you test them together at one time? By testing each one individually you reduced the load on the IMC.

 

[Quattro]

Hi,

Thanks for the reply.
Yes the processor has been overclocked for last 2 years but since I began troubleshooting I have reverted to stock values.

I don't know how to make memtest86 test each stick individually while both sticks inserted. I'll have to have a look at the settings in memtest86 and try that.
If I needed more voltage then wouldn't I have always needed more voltage? What would cause it to require more now when it's been previously entirely stable?

 

[trents]

Hi,

Thanks for the reply.
Yes the processor has been overclocked for last 2 years but since I began troubleshooting I have reverted to stock values.

I don't know how to make memtest86 test each stick individually while both sticks inserted. I'll have to have a look at the settings in memtest86 and try that.

You misunderstood me. I'm saying to not test them individually but to test them together. This would not be testing the RAM itself so much as it would be testing the effect of stress on the Integrated Memory Controller. That component lives in the CPU and is responsible for handling the data that flows between the cores and the cash and the RAM. Two sticks will put more load on it than just one. What testing you have done already has proved that the RAM is good but has not proved that the IMC can handle the greater load of two sticks of RAM as opposed to just one.

If I needed more voltage then wouldn't I have always needed more voltage? What would cause it to require more now when it's been previously entirely stable? Because silicon degrades over time and it degrades more quickly when components are overclocked.

 

[Quattro]

Yes I have tested both sticks, as per my original post "I tested them individually in Memtest86 for four passes, no errors. Tested them together in slots 2 and 4, no errors. Tests done with and without XMP."

Not sure what to try next.

 

[Zerileous]

It is also possible that your silicon has degraded the point where stock voltage is not adequate for stock clocks. You can try to run stock clocks with a positive offset voltage, or even the voltage you had been using for your OC.

I don't think memtest86 is going to show the whole picture, as it doesn't stress the memory, or the IMC. Repeating the one stick at a time test using a stress test like AIDA64 or Prime94 (set to blend and then check custom and set 28GB memory use) will tell us more.

Have you moved or bumped the case around that time? I would almost suggest a teardown, deep cleaning and re seating everything including the CPU if none of the above yield answers.

 

[Quattro]

I'll try those stress tests as you suggest and report back tonight. I haven't torn down the build since I put it together, it's possible it just needs a proper clean. I do live in a very dusty climate. The RGB died on my Kraken cooler months and months ago so maybe I should replace that anyway in case something else is going wrong with it, I absolutely hate the software for it anyway.

 

[trents]

It is also possible that your silicon has degraded the point where stock voltage is not adequate for stock clocks. You can try to run stock clocks with a positive offset voltage, or even the voltage you had been using for your OC.

I don't think memtest86 is going to show the whole picture, as it doesn't stress the memory, or the IMC. Repeating the one stick at a time test using a stress test like AIDA64 or Prime94 (set to blend and then check custom and set 28GB memory use) will tell us more.

Have you moved or bumped the case around that time? I would almost suggest a teardown, deep cleaning and re seating everything including the CPU if none of the above yield answers.

Not sure I agree with that. My fans ramp up some when run memtest86. It must put some stress on those components. I don't see how it could be checking the memory without making it do some work with data.

But I still thing the OP should try giving the CPU and the IMC some more juice. He himself states he is only using 1.28 volts for an overclock of 5 ghz. That's probably on the low side and worked until some deterioration set in. I mean a lot of people are in the 1.35 volt range in order to get stable at 5 ghz.

 

[Quattro]

Tore the whole thing down, removed motherboard from case and reseated everything inc processor after cleaning off the IHS and putting new thermal paste. Compressed air to clean the rad and the vrm coolers etc. Wasn't much dust at all.

Put it back together again and just for interest ran timespy stress test as I was curious what temps my cpu/motherboard would be, anyway it ran 20 loops without BSOD. So I played a game that was always generate a bsod before, nearly an hour and no crash or bsod so applied previous overclocks to cpu and gfx card. Another hour and no bsod at all.

Maybe reseating everything and blowing out the cobwebs actually fixed it! I'll test some more tonight just to be sure that it's all ok.

 

[habbajabba]

I had pretty much the same problems on my rig after about 5 yrs running at 4.5ghz oc from the 3.7 stock. I'm running @stock right now but need to locate my paste to reseat the aoi cooler block again. i thought it was the small msata hd I replaced but with the heat put through the thermal paste it's probably just good practice to replace now and then. When I cracked it it kinda looked old and not too fluid anymore.

 

[Quattro]

Good to know this forum is still here and active. Thanks for your help gentlemen.