Solved BSOD possibly graphics card or memory related

[mortikaye]

Hello.

I've been getting BSODs over the last few weeks with differing messages. Bad pool header, memory management, nvlddmkm.sys and others.

I've had the ram and the graphics cards out and used compressed air to clean the cards and the slots.

I've flashed my bios to a more recent version.

I've ran malwarebytes.

I've removed the graphics driver with DDU and installed a different driver.

Ran memtest and it came back with no errors.

I've done a non destructive reinstall of windows.

I've tried all the above, but am still getting BSODS.

 

[Boozad]

There are no dumps in your logs.
There aren't any .dmp files in your attachment. These are required to analyse the BSODs.

Follow these steps:

1. Configure Windows to create minidumps:
http://www.sevenforums.com/tutorials/174459-dump-files-configure-windows-create-bsod.html

2. Set the paging file on C: to be the same size as the physically installed RAM:
Virtual Memory Paging File - Change

3. Set the Windows Error Reporting service to MANUAL:
http://www.sevenforums.com/tutorials/2495-services-start-disable.html

4. Test to see that dump files are created:
- Open Windows Task Manager
- Navigate to the Processes tab
- Right-click on explorer.exe, and click Create Dump File

- Note the location of the saved file

- Check that location and confirm that explorer.dmp exists

 

[mortikaye]

I don't understand why it's not creating dumps.

Write an event to the system log was checked.

What am I doing wrong?

 

[Boozad]

I don't know, have you followed all of the steps above?

 

[mortikaye]

OK, I've done the above and explorer.dmp exists

 

[Boozad]

Next time you receive a BSOD run the Log Collector again and upload the resulting .zip file.

 

[mortikaye]

Will do.

Thank you.

 

[Boozad]

You're welcome.

 

[mortikaye]

Annoyingly my pc hasn't crashed with a BSOD the last day or so. Now it's taken to freezing for 15-20 seconds every minute or so.

Very irritating.

 

[mortikaye]

I've re-run the log collector and hopefully it has the dump files attached in the zip.

Thanks for taking a look and sorry it's taken me this long to get back to this.

 

[Boozad]

Hello mortikaye,

Right, there are four dumps in your logs with the latest two (0xFC and 0x101) blaming your nVidia driver. The older 0x101 was inconclusive.

fffff880`147ea650  fffff880`05f99f00Unable to load image \SystemRoot\system32\DRIVERS\nvlddmkm.sys, Win32 error 0n2
[COLOR=Red]*** WARNING: Unable to verify timestamp for nvlddmkm.sys
*** ERROR: Module load completed but symbols could not be loaded for nvlddmkm.sys[/COLOR]
 nvlddmkm+0x76ef00

6: kd> lmvm nvlddmkm
start             end                 module name
fffff880`0582b000 fffff880`062e6000   nvlddmkm T (no symbols)           
    Loaded symbol image file: nvlddmkm.sys
    Image path: \SystemRoot\system32\DRIVERS\nvlddmkm.sys
    Image name: nvlddmkm.sys
    Timestamp:        [COLOR=Red]Tue May 12 02:59:00 2015[/COLOR] (55515E64)
    CheckSum:         00A86612
    ImageSize:        00ABB000
    Translations:     0000.04b0 0000.04e4 0409.04b0 0409.04e4

You also received a 0x124 just over two weeks ago, a 0x124 denotes a hardware error. These hardly ever give an indication as to what the problem is and that's the case here, although they're usually fatal errors whereas this one isn't.

FAILURE_BUCKET_ID:  X64_0x124_GenuineIntel__UNKNOWN

BUCKET_ID:  X64_0x124_GenuineIntel__UNKNOWN

ANALYSIS_SOURCE:  KM

FAILURE_ID_HASH_STRING:  km:x64_0x124_genuineintel__unknown

FAILURE_ID_HASH:  {261c7ae5-2cd4-7cad-992f-2df21b8bc48b}

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

4: kd> !errrec fffffa800f99f028
===============================================================================
Common Platform Error Record @ fffffa800f99f028
-------------------------------------------------------------------------------
Signature     : *** INVALID ***
Revision      : 255.255
Record Id     : aaaaaaac00000000
Severity      : Recoverable (0)
Length        : -1431655764
Creator       : {00000000-aaac-aaaa-8000-000000000000}
Notify Type   : {00000000-aaac-aaaa-8000-000000000000}
Flags         : 0x00000080

What I would like you to do is remove one of your GPUs from your SLi setup and test the rig with just one card. Make a note of which driver you're using and temperatures at both idle and under load. If things are stable with just one card, replace it with the second card and test the system again.

Report back with the results and we'll work from there.

 

[mortikaye]

thanks for the quick reply.

I was planning on using Open Hardware Monitor for measuring the temps. Would this program be OK, or would you suggest another?

 

[Boozad]

There are a few you can choose from, I usually use Speccy (linked in my signature) or HWMonitor is another I've used. I've no experience of Open Hardware Monitor so I couldn't say, you can install all three if you like which should be helpful in determining the readings are correct.

 

[mortikaye]

For the first card I tested it was 28°c at idle and was 69° at load.

Things seemed stable. I played the Witcher 3 for a few hours with no CTD or BSOD.

Will test the other card tomorrow and get back to you with results.

Thanks again for looking into this.

 

[Boozad]

No problem at all mate. Just keep us informed.

 

[mortikaye]

I did at proper stress test with FurMark on that first card for about 15 minutes and the it got up to about 81°c, but no problems. No CTD or BSOD.

Will stress the 2nd card tonight.

 

[mortikaye]

2nd card idled at 28°c and at load got up to 85°c.

Things seem stable. No CTD or BSOD.

 

[Boozad]

Is the original card in the first PCI-e slot? Install the cards in the opposite slots to what they were originally and set up the SLi configuration again, making sure you perform a clean install of the drivers. Test the system again and see what happens.

 

[mortikaye]

Sorry it's taken me some time to get back, but I've just moved house.

OK, so I did as suggested and set up the sli again, but was still having issues. I then put a different sli bridge in and the problems seem to have gone.

 

[Boozad]

Excellent news. If you're happy the issue is resolved please mark the thread as Solved, you can always mark it Unsolved if the issues return. I'm glad you've found the root of your problem.