Ongoing BSOD; dxgkrnl.sys, dxgmms1.sys, and nvlddmkm.sys

Terrek · 05 Jan 2017

Hi folks, I'm having intermittent issues with my computer. It started several months ago, and I thought I had stopped it by putting in a different GPU, (which was older), but it started up again.

Since my original posting I have installed a new GTX 1060 FTW card and a new EVGA SuperNOVA 850 G2 power supply.
Original thread here: BSOD possibly from Nvidia nvlddmkm.sys Driver

I am mainly getting BSOD when waking the computer up from sleep mode. It seems to happen maybe once every week or so, sometimes more, for no apparent reason.

BSOD is still being caused by dxgkrnl.sys+5d1f0, although dxgmms1.sys+37f54 and nvlddmkm.sys+9584a0 are still flagged in red with BlueScreenView.

I ran Memtest for twelve hours straight about a week ago, and it detected no issues with my memory cards.

At this point only the motherboard, cpu, hard drive, and memory have not been replaced, but I'm not sure which is the culprit. They are all around 5 years old.
I would guess the memory would be most likely, but would like to hear your thoughts before I spend even more money on this damned issue.

attached is most recent bsod

ICIT2LOL · 11 Jan 2017

Hello Terrek mate just having a quick look at the dumps there is a graphics issue with the card driver

Code:

*** WARNING: Unable to verify timestamp for nvlddmkm.sys
*** ERROR: Module load completed but symbols could not be loaded for nvlddmkm.sys

Now you could try redoing the driver for the card first to see if the problem persists EVGA - Download Center now I am jot sure if this is the right driver for the device you have an the nearest I could get was for an EVGA card ??

Now as for the memory I don't know what you used for the test and if it was the Windows feature then it is pretty weak.
Now to make absolutely we are on the same page the memtest linked here is the*only*one we recommend and it MUST be run for at least 8 passes or you may as well not bother. It is rather lengthy and best left for a downtime for that machine - overnight is good. Having said that you can stop the test if errors show up while you are around and then it is a matter of tracking down the bad stick of RAM.*
RAM - Test with Memtest86+

Lets get those out of the way first. Plus as you installed a new card did you think about cleaning the slot before installing? If not then this is what I do
SLOT CLEANING
Now as for the slots to clean them use a strip of old credit card (or any card thin enough some are better than others ) about 1/2 inch wide nip the corners off one end - preferably round them off hold a piece of thin lint free cloth over the end dab on some form of alcohol - isopropyl alcohol (what we swab the skin with before an injection) is good and GENTLY swipe along the slots.
This can also be done for the GPU slot/s and also the contacts of the sticks or cards - not the stick but some lint free cloth of course. See my pic for the sick but please be very gentle doing it if you are in any doubt - please ask before doing it but I use this all the time and have yet to damage anything.

Edit: I mention cleaning the slot because of the age of the board and it may likely be that crud has built up down in amongst the slot contacts.

Terrek · 11 Jan 2017

Hello

It was Memtest86+ that I used, and like I said, I ran it overnight for over 12 hours. I think it completed between 10 and 12 passes, don't remember for sure, but it didn't flag any problems.

As for the video card, I used canned air to clean the slot before installing the card, and did a clean install when I installed the drivers, and again when I updated them shortly afterward.
I'm not really comfortable sticking anything in the slots though, I am quite paranoid when it comes to anything that could potentially damage my rig.

ICIT2LOL · 11 Jan 2017

Ok that's good then and the slots you do whatever you feel comfortable with.

Now I guess as there is a reference to the graphics driver the next step would be to update the driver for the card Drivers | GeForce

The dxgkrnl.sys is listed in the dumps chart as relating to Windows DirectX Grpahics Kernel and is notated as a Windows update issue. I am not sure of how to deal with this one as quite a few dumps have update issues attributed to them.

So it might be worth a try with the driver first.

Terrek · 21 Jan 2017

well its been over 2 weeks since last BSOD, and it did it again two days ago. I immediately ran scf/scannow in command prompt, no issues detected. I then performed clean install of my NVidia drivers.

Everything was fine yesterday, but just now when waking up my pc I got the same BSOD.

I have noticed my pc seems to be running a bit slower lately, especially when playing my new games like Just Cause 3 and Fallout 4.

I ordered new memory that I have been eyeing for awhile since they are on sale. This will effectively double my current memory from 8gb to 16gb
G.SKILL Ripjaws X Series 16GB (2 x 8GB) 240-Pin DDR3 SDRAM DDR3 1600 (PC3 12800) Desktop Memory Model F3-1600C9D-16GXM.

I'm hoping this helps, the only other components not yet replaced are motherboard, hard drive, and processor. Otherwise I'm thinking maybe something is messed up with direct x, since it is flagged twice

Terrek · 21 Jan 2017

Ok here are some screenshots from running Memtest-86+

First is from scan I did back in December without changing any settings. As it turns out, it ran for over 8 hours, and completed 6 passes, so I was incorrect in my earlier post.

The second one is from last night. Not sure if I did something wrong, but I went under Configuration, then Test Select, then chose Default. I let it run for six hours and it came up with the 25186560 errors. However, the pass % and the time stamp never changed, so I wonder if it kinda locked up, even though the number of errors kept climbing.

I then went under Configuration. then Error Report, then Error Summary and let that run for 2 hours.
All the stuff in red disappeared, but the error numbers were still there. The pass number jumped up to 12.
It never showed the Memory SPD Information like in the first photo, so I'm not sure if the cards have really gone bad or not.

I'll let it run again tonight without touching any settings again and see what it does.

ICIT2LOL · 21 Jan 2017

Good idea mate as I am pretty sure the RAM is corrupt in some way. I know it is tedious but worth it in the end if it sorts out the problems the machine has.

Terrek · 22 Jan 2017

well, I ran Memtest again for over 12 hours and 9 passes, no errors detected....So I don't know what's going on. My guess is either one of the cards or one of the slots is intermittently failing?

I'm just gonna test the new cards when they get here one at a time to make sure that they and the motherboard slots are good. I need more memory anyway so I'm not really concerned about whether or not the old cards are good at this point, as long as they last until the new cards get delivered in the next few days

ICIT2LOL · 22 Jan 2017

Ok mate now all you need is one of the good sticks and test each slot. It is very tedious but if there is a failing slot then this should pick it up.

So just how many channels are you using right now? Now if it were me I would up the ante to 16GB and use a matched set of 4x4GB sticks that way it uses all of the channels and spreads the laod as it were.

I may have this wrong and stand to be corrected but generally speaking it best using more sticks to get the most of the available channels and RAM sticks. For example using 1x8GB stick would use only one channel and all data being fed to and from the CPU has to utilise the total memory on that one stick, as opposed to 2x4GB sticks would use two channels and the data load spread over the two sticks if that makes sense.

Terrek · 23 Jan 2017

I currently have 2x4gb sticks, upgrading to 2x8gb sticks.

As I said, I'm just going to test the new sticks and the motherboards slots at the same time. I needed to upgrade my memory anyway, so I'm basically considering the old sticks as garbage, whether they are corrupted or not. They are over 5 years old, so basically past their life span imo