Multiple Daily freeze and now BSOD- Link to V2 log and dump files

Problem: Windows 10 machine well established started to freeze. No mouse movement. Have to hard reset to get computer to interact. Recently there have been BSODs now too and event viewer entries for:

Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 26

Update your browser to use Google Drive, Docs, Sheets, Sites, Slides, and Forms - Google Drive Help

What I have done-
-Clean boot
-New NVME HDD with fresh install windows 10 and fresh NVidia, AMD, etc drivers
-Bios update
-Memtest86+ 40 passes with no errors
-OCCT stress tests of video, CPU, power with no error
-Reseated video card and all MB connectors
-At one point in past I heard fans surging but no longer, this maybe because I chainged fan setting not up to perormance
-No overclock on CPU ever but I have operated with xmp on and then off. Doesnt seem to affect.
-All hardware relatively new except GPU.

Can you upload the full kernel dump please? It's the file C:\Windows\Memory.dmp. This particular bugcheck can only be fully diagnosed with the kernel dump.

I suspect the troublesome driver is likely to be nvlddmkm.sys, the graphics driver (or the graphics card). There are a lot of errors in your log for nvlddmkm.sys, but we need the full kernel dump to be sure.

Thanks I have seen you help many people while researching this issue of mine and I am thrilled you are engaging here. Thanks for what you do.

In the google drive link above have the memory dump file as it existed after 3 of last BSODs along with minidump.

Are you able to see in that link?

I will try to upload the 1 g file too on side in a new post but so far website pausing and not completing on it.

Thanks again!

Ah, sorry. I only downloaded the V2 log file. I'll download them all and get back to you.

ubuysa said:

Ah, sorry. I only downloaded the V2 log file. I'll download them all and get back to you.

Thank you VERY much!

The way we tackle this particular BSOD is a bit different. The BSOD happened because a collection of DPCs (the back-end of device interrupt processing) collectively ran for too long. To find the biggest contributors for this we need to extract the WMI trace records from a kernel dump, export them in a form that the Windows Performance Analyzer can read, and then use WPA to look at the list of DPCs that were running. Here's the output for your kernel dump...



The Duration Fragmented (ms) sum column is the one we need, it's sorted on DPC run time so the longest running DPCs are at the top (note that ntoskrnl.exe is not a DPC, it's the Windows kernel, so we ignore that). Microsoft recommend that no DPC run for longer that 100 microseconds (0.1ms) and you have several that run longer - some considerably longer. They fall into two groups however; networking (tcpip.sys and ndis.sys) and graphics (dxgkrl.sys and nvlddmkm.sys). In addition, Wdf0100.sys (the Windows Driver Foundation high-level driver) and HDAUdBus.sys (the Windows high-level audio bus driver) also run for longer than recommended but I think we can ignore these as the others clearly contributed much more to the cumulative overrun (nvlddmkm.sys in particular).

It would seem that a streaming operation was in progress here? That explains why we see graphics DPCs and networking DPCs running (and an audio DPC as well). On the face of it the Nvidia graphics driver (nvlddmkm.sys) contributed most to the cumulative overrrun, but that driver is fairly current (but do check for an update). I wonder whether the problem here might not be network related? A slow running networking DPC might also impact the graphics DPCs that depend on them (when streaming).

The tcpip.sys driver is a Windows high-level driver, it will be calling the third-party network adapter driver to access the network. You're using your Intel I211 LAN adapter and its driver is e1r68x64.sys. The version of this driver that you have installed looks to be quite old, dating from Feb 2021...

Code:

23: kd> lmDvme1r68x64
Browse full module list
start             end                 module name
fffff806`81d50000 fffff806`81de3000   e1r68x64   (deferred)             
    Image path: \SystemRoot\System32\drivers\e1r68x64.sys
    Image name: e1r68x64.sys
    Browse all global symbols  functions  data
    Timestamp:        Tue Feb 23 15:42:01 2021 (60350629)
    CheckSum:         0009E4BA
    ImageSize:        00093000
    Translations:     0000.04b0 0000.04e4 0409.04b0 0409.04e4
    Information from resource tables:

I would suggest that you download the Intel Driver & Support Assistant and use that to look for updates for your Intel devices (especially that LAN adapter). Install any drivers that it finds.

The version of nvlddmkm.sys that you have is much more recent...

Code:

23: kd> lmDvmnvlddmkm
Browse full module list
start             end                 module name
fffff806`90530000 fffff806`93e49000   nvlddmkm   (deferred)             
    Image path: \SystemRoot\System32\DriverStore\FileRepository\nv_dispi.inf_amd64_1fea8972dc2f0a69\nvlddmkm.sys
    Image name: nvlddmkm.sys
    Browse all global symbols  functions  data
    Timestamp:        Thu Nov  9 23:35:30 2023 (654D50A2)
    CheckSum:         03804E98
    ImageSize:        03919000
    Translations:     0000.04b0 0000.04e4 0409.04b0 0409.04e4
    Information from resource tables:

However, there is an updated driver dated 4th Dec 2023 (546.29) for your GTX970 on the Nvidia driver download site. I suggest you install that updated driver.

The problem of course could be hardware, it might be your graphics card or even the LAN adapter. If it still BSODs with a DPC_WATCHDOG_VIOLATION please run another V2 log and upload the output, also upload the kernel dump (C:\Windows\Memory.dmp).

Status update-

I was working through suggestions from @ubuysa

-to remove NVidia drivers in prep for driver updates as suggested I used DDU which suggests safe boot and several restarts in between cleaning steps. I remoted both display and audio drivers as both are options in DDU with network unplugged

-Running out of time I got to a clean step and rebooted into safe mode and only a single screen active and low res. Safe mode. Has to stay that way since off to work.

-After 24 hrs no crashes, no event log entries suggesting any restarts. Recently I cannot make 24hrs with out BSOD or cache exception restart.

-planning to proceed with the NVidia install I rebooted into normal boot and before I could start driver update process for NVidia (while I was still on Microsoft display adapter)....I got restart and event log said cahce error again.

-In notices fans surging on PC so I opened OCCT and grabbed screenshots. I attached that and event log entry for WHEA 18 that I got causing the restart.

-At this point I turned on network so as to write this post and windows added NVidia driver so I will go back through DDU steps and hen once clean install downloads drivers from NVidia site.

-After that I will start into steps proposed by @zbook


Does the stability in safe mode suggest hardware is OK or can it still be hardware?

Does the core voltages in attached WordPad file suggest a problem given core voltages during this "surging period" seem to get pretty low?

I add pictures of OCCT at this link along with new V2 log collection under folder with today's date here:

Update your browser to use Google Drive, Docs, Sheets, Sites, Slides, and Forms - Google Drive Help

The Safe Mode success suggests that it's not hardware but a software/driver problem. The recent dumps are 0x133 with argument set to 1, indicating an issue with several driver DPCs as before. I would now suggest you perform a Clean Boot of Windows. This loads a full Windows system and if you disable all third-party drivers as suggested in the link, the system should should be stable - though probably not very useable.

Now you can use a boolean search technique to find out which driver or drivers cause the BSODs. Clean Boot again and enable half the third-party drivers. If it BSODs the flaky drivers are in the half you enabled. If it's stable the flaky drivers are in the half you didn't enable. Keep halving the drivers in this way to quickly home in on the one(s) that cause the BSODs.