Multiple BSOD's, I'm at a Loss

[chrls.ocnnr]

Hey guys,
I just updated my wife's mobo, cpu, and ram.

Mobo gigabyte Z370 hd3

Cpu 8600k

Ram corsair vengeance lpx 2400 16GB

At the same time, I upgraded my mobo and cpu to the same, except I got g skillz (note I'm not having any issues)

Now her comp is completely flipping out. BSOD for critical structural failure, memory management, ntfs.sys, page fault in nonpaging area, dxgkrnl.sys

She tried a fresh install of windows 10 pro, then said screw it and tried to install sparky Linux gameover edition, then when that wouldn't install (freeze while unsquashing image file) she went back to fresh install of windows 10 pro.

I've ran the scannow, ran all the dism checks and restores, ran disk checks in admin command prompt, ran windows mem test, reset bios, and updated all the drivers (I think). Only test I haven't ran is the official memtest everyone talks about (will run it tomorrow).

Note: it would bsod within seconds of opening a game. Then I turned on xmp profile 1 in bios, and that seemed to stabilize games for awhile. She was able to play for about 30 mins, then a unity crash. Another 30 mins, then a unity crash and back to BSOD critical structure failure.

Also note: no overclock at this time. I did a minimal overclock for her right when we set everything up the first time (45, it boosts itself to 43 already) and it bsod'd. I figured the voltage wasn't high enough for the oc or something, so I just reset defaults.

I really thought I had fixed everything after turning on the xmp profile. Left the computer on for a couple of hours after briefly testing a game and it was fine. I'm totally at a loss now.

From my reading, all signs point to ram. Given the crazy amount of different errors I'm getting though, what do you guys think? Could it be that I just need to get another windows key off kinguin, or something else? Any and all help is appreciated, thanks for taking the time to read this.

 

[Colif]

dxgkrnl.sys points at 1 cause being the GPU drivers, I can't guess what the others might be caused by. Memory management can be drivers or ram,

Can you follow option one on the following link - here
and then do this step below: Small memory dumps - Have Windows Create a Small Memory Dump (Minidump) on BSOD

that creates a file in c windows/minidump after the next BSOD
copy that file to documents
upload the copy from documents to a cloud server and share the link here and I will get someone to convert file into a format I can read

Have you run memtest86 on the ram sticks? Check one stick at a time, up to 8 passes. Only error count you want is 0, anything higher is likely cause of the BSOD. Any ram sticks with errors need to be removed/replaced. IT makes a bootable USB so no need for an operating system.

 

[chrls.ocnnr]

Thanks Colif, I'll upload the minidump this afternoon after work using the instructions.

Haven't tested the ram that way yet. I'm not really sure how, but I know there's a bunch of guides on it. I'll do that this evening as well. Thanks for the quick response

 

[Colif]

Is XMP on? MIght want to run memtest at stock speeds

I haven't done it myself but I believe the memtest download creates the program itself on a USB and you just boot off it
How to boot off USB without messing around changing boot order in BIOS - If you in windows, you can go to settings/update & security/recovery
under advanced startup, click restart now button
this restarts PC in advanced startup
on this screen, there should be an option called Use device.
put Memtest86 USB into PC and choose it from list
PC will restart and run from USB (might be faster in long run to change boot order as will need to do this each time you swap ram sticks, its your choice)

Run it on 1 stick at a time, so remove all sticks except the one being tested. Up to 8 passes, but if you get errors right away its not really necessary to do 8 passes. Check all the ram this way, it can take a while though.
Only score you want is 0

 

[chrls.ocnnr]

Alright Colif,
I ran the test 8 passes with both sticks (I know you're supposed to just do it once, but if it's the RAM I'm gonna replace both anyway, so hopefully that won't affect the outcome.

I got 1 error on the 8th pass. It says # Tests Passed: 47/48 (97%)

Lowest Error Address : 0x39B453F8C (14772MB)
Highest Error Address: 0X3F72C600C (16242MB)

Bits in Error Mask: 0000000000002000
Bits in Error - Total: 1 Min: 0 Max: 1 Avg: 1
Max Contiguous Errors: 1

Still haven't done the minidump file thing yet. Can you gather anything from the information I just posted?

 

[nobspls]

If you even get one memtest failure it probably means the memory is unreliable. I had memory test that ran 72 hours before the first failure. And what it lead to was random BSOD once a couple of weeks. Sometime it might even got a month. But when you can't trust that memory it is just a no go. I would strong push for replacing/refunding/exchanging that memory asap while you still got your return policy time window.

Trying to downclock, not use that XMP profile etc. only hides the issue, and the delays will make it so that your fall out of your return policy time limits. It is a hassle, but it is a lot more hassle to try to work around it and then suffer the consequence later. I am not impressed at all with DDR4 reliability. I haven't tally too closely, but I had to return and exchange at least 3 sets of DDR4 in the past 12 months for various system, probably worked on a dozen or so machines, so that is pretty high failure rate of 25% or so odds. I really hate to do memtest for 72 hours straight, but it is now the burn in standard because I can't trust the DDR4s I am getting.

 

[chrls.ocnnr]

Thanks for the reply nobs. I think I'm outside my return window already. Anyway, I'd rather just be done with corsair and get g skillz for her. I'll try to return the old though.

Just to be sure, would it be a good idea to put my ram in her system to see if that changes anything? Or is that not necessary? Thanks

 

[nobspls]

That would be a good test to verify it is indeed the memory. Because you can then run the test on your machine and should see similar errors. It is always better to take that kind of hassle punishment yourself, then being yelled at about a machine that isn't working.

BTW I am using Corsair Vengeance LPX DDR4 3000 Cas15 2x8GB. I got 2 pair of these for 2 different machines, and I had to exchange one pair of those in its first week when I was doing the burn in testing. It was after like 48 hours that memtest86 was showing errors.

Here is a technique that may work for you to return stuff out of the return period. But you will need to make sure they do not track serial numbers, and it will require you double down on the memory you got. Basically you buy another set of same stuff and then return the flaky one as the one you got. Fundamentally you are not stealing, and the arbitrary return period these sellers impose is really unjustified, besides doesn't Corsair have lifetime warranty on these things anyways. All this really means is that the seller will have to deal with that hassle but they probably have an easier channel to Corsair anyways. It is an ethical gray area, but considering how these manufacturers/vendors didn't care two cents about ethics when they cut corners on quality control, I would not fault you for sticking it to the man.

 

[chrls.ocnnr]

Well, I popped one of my ram sticks in her rig and she hasn't bsod'd for about 2 hours (was happening every 5-20 mins consistently). Went ahead and ordered her a pair of the same I have. I would have no problem doing the ol bait and switch, I'm just honestly not interested in using corsair ram now lol. Hopefully I can return it or something, but if not oh well, live and learn. I really appreciate everyone's help, and hopefully I won't have to come back in a day or two with anymore issues. You guys are the best, thanks again!