Infrequent BSOD becoming more frequent

BizzaroHunter · 26 Feb 2016

[EDIT] I see now I posted this in the wrong area of the forums. SORRY! If someone could move this to the BSOD section I would appreciate it.[/EDIT]

I built this PC three, maybe four, months ago now and haven't had many issues with it after initial troubleshoot on the build. Had to use some old hardware in this rig until I can get enough money to finish it. But now, my (maybe) once a month BSOD is becoming a more regular thing.
Attachment 66425
As you can see, it started out rarely enough but it's getting way, way more frequent. I began further investigation today at around 4am and found a few tools I didn't know existed, including SFC and DISM. SFC returns that I have a corrupt file, and it is irreparable. "opencl.dll" is the file. I found a download on dll-files and tried replacing it manually, but that does nothing for the SFC scan, and I've had BSOD since. I downloaded Win10 TechBench from Microsoft and mounted it, and used option 3 from image_health.cmd. While unmounting, I got BSOD again. As I type this, my SFC /scannow finished, found 1 file(opencl.dll), and replaced it from store properly this time.

Since I can't really duplicate the functions that caused the BSOD, I'm wondering if anyone has any other advice for me? I'd like to be able to check this under 'solved'.

I am attaching the Dump Collector zip. As far as what I'm doing when these BSOD's occur - it ranges from playing a video game, watching videos on YouTube, reading through forums, or even unmounting an .iso as it did just a couple hours ago.

Arc · 26 Feb 2016

Your motherboard manufacturer says ......

Code:

- Supports DDR4 3000+(OC)*/2933(OC)/2800(OC)/2400(OC)/2133 non-ECC, un-buffered memor

Your RAM modules are ....

Code:

  Size                          4096MB
  Speed                         2666MHz

So probably there is a mismatch. Rectify it, See how it goes. Let us know.

The BSODs are due to memory corruption.
______________________________________________________________________

Code:

BugCheck A, {e38, 2, 0, fffff801be4e22c7}

Probably caused by : memory_corruption ( nt!MiUnlinkPageFromList+87 )

Followup:     MachineOwner
---------

Code:

BugCheck 3B, {c0000005, fffff8013d8c8eaf, ffffd00022539670, 0}

Probably caused by : memory_corruption ( nt!MiIdentifyPfn+1ef )

Followup:     MachineOwner
---------

BizzaroHunter · 26 Feb 2016

Arc said:

Your motherboard manufacturer says ......

Code:

- Supports DDR4 3000+(OC)*/2933(OC)/2800(OC)/2400(OC)/2133 non-ECC, un-buffered memor

Your RAM modules are ....

Code:

  Size                          4096MB
  Speed                         2666MHz

So probably there is a mismatch. Rectify it, See how it goes. Let us know.

The BSODs are due to memory corruption.
______________________________________________________________________

Code:

BugCheck A, {e38, 2, 0, fffff801be4e22c7}

Probably caused by : memory_corruption ( nt!MiUnlinkPageFromList+87 )

Followup:     MachineOwner
---------

Code:

BugCheck 3B, {c0000005, fffff8013d8c8eaf, ffffd00022539670, 0}

Probably caused by : memory_corruption ( nt!MiIdentifyPfn+1ef )

Followup:     MachineOwner
---------

I thought this might have been an issue as well so I originally ran a memtest when I got the system, and it made 4 passes with no issues.

I checked on the mobo manufacterer's website(ASRock > Fatal1ty X99X Killer) and the model number for the memory(G.SKILL Ripjaws 4 Series 16GB Desktop Memory Model F4-2666C15Q-16GRR - Newegg.com) is listed under the supported memories. Is it possible one of the sticks has gone bad and it is causing this memory corruption issue? I'll run memtest again as soon as I post this and edit with the results.

UPDATE: I've finished two passes and I'm letting it do a couple more as I post this update. First pass, 0 errors. Second pass,2 errors in test 9.

Arc · 27 Feb 2016

BizzaroHunter said:

UPDATE: I've finished two passes and I'm letting it do a couple more as I post this update. First pass, 0 errors. Second pass,2 errors in test 9.

Errors/red lines means one or more RAM is faulty. But the fault may occur due to a faulty DIMM slot, too, which is a motherboard component. Using memtest86+, you can discriminate between a faulty RAM and a faulty motherboard.

How? Say you have two RAM sticks and two DIMM slots. You obtained errors at the test with all RAM sticks installed. Now, remove all the sticks but one. Test it in all the available slots, one by one. Continue the same procedure for all the available sticks.
How to make the inference that is it a RAM issue or it is a motherboard issue? Suppose you have got the result like that:

Attachment 66590

It is a RAM, a bad RAM.

But if you have got a result like that:

Attachment 66591

It is a motherboard issue. The particular slot is bad.

If it is a bad RAM, you need to change the RAM. Take the other existing (good) RAM to the shop and get one of the exactly same specification. The best is, get a totally new RAM KIT.

If it is a bad slot, it is a motherboard issue. But changing the motherboard needs a preparation for most of us; so dont populate that slot for the time being.

BizzaroHunter · 28 Feb 2016

So I need to individually test each stick in each slot? 4 sticks, 4 slots, 16 tests. And how many passes do you recommend I do?

As mentioned before, I received only 1 error in 2 passes. So how can I be sure that I'm not getting a false positive with any test? I let memtest run longer than that, but either my roommate pressed escape before I could see the results or the system crashed and restarted.

Arc · 29 Feb 2016

Apparently thee is no false positive for memtest86+.

One single error is enough to rule it out.

And yes, if you want to go to the root of the issue, you have to do it.

About passes .... 8 passes or appearance of one issue, whichever is earlier.

BizzaroHunter · 29 Feb 2016

I've been doing tests per stick of memory, per slot, and I think I've narrowed it down to a bad stick of memory.

However, thus far I've completed 6 passes with this memory in my B1 slot with 0 errors. Is it within the realm of possibility that it could be some sort of mismatch within that one slot, with that one stick of memory?

Arc · 01 Mar 2016

Read the post #4 once again, you will get the answer.