Win10 BSOD problems

[Nagisan]

Recently rebuilt my computer (replaced most parts of it) and everything seemingly worked fine. I successfully ran memtest for 8 hours, and some various stress tests to ensure cooling wasn't an issue. However, since running the tests I seem to get crashes every now and then. I don't often reboot my computer (instead utilizing Sleep mode) so that may have something to do with it, but my old system had no problem running for weeks going in and out of sleep with no issues on Windows 7.

 

Below are memory dumps of the last two crashes (I had more but CCleaner was accidentally set to clean up old memory dumps and got rid of three of them before I realized this).

 

I've done what I know, looking at the dumps with WinDBG, but they both point to ntkrnlmp.exe causing the fault which, to my understanding, is usually caused by something else and in fact not ntkrnlmp.exe. Unfortunately I do not have enough experience with BSODs to diagnose this any further and feel like it is not a memory issue given the 8 hour test, and not a processor issue as it also has passed some stress testing but again, not familiar enough with the issue to continue.

 

If it makes any difference, the last crash in particular happened when I was simple changing tabs in Chrome.

 

https://www.dropbox.com/s/v7jfmbc0rk5dbzr/070916-5687-01.dmp?dl=0

https://www.dropbox.com/s/ksetqsbgus0otby/071616-4906-01.dmp?dl=0

[Danielx64]

When you rebuilt/upgraded your computer did you do a clean install? Are your drivers upto date?

[Nagisan]

Yes, clean install of Win10 (was running Win7 on my old system). Drivers are all the latest I can find on the manufacturers website for everything.

[noneofyourbusiness]

Even though your memtest says the memory is good, I would still recommend testing your memory.

Take most of the memory out, if can just leave 1 then do. Run it for awhile. Do the errors pop up? If no, run the next memory. Do for all.

Sometimes memtest says good, but when memory actually used, one can show up as bad and cause the errors you refer to.

[Nagisan]

How long should I run the memory test on each stick (2x8GB)? I've heard some people do crazy things and run it for 12+ hours, but, especially if running one at a time, wouldn't 6-8 hours overnight be enough for each?

[techbeck]

Download this.   Will give better info on the BSOD.

 

http://www.nirsoft.net/utils/blue_screen_view.html

[Nagisan]

I tried that before I grabbed WinDBG but was less familiar with it so I wasn't sure if there was a way to get any more info out of it.

 

It actually shows almost the exact same info, just easier to read.

 

If it helps, screenshots of the properties of each minidump as seen by Blue Screen View are now attached.

[xendrome]

yeah being NTOSKRNL it's likely a memory or voltage/power issue, I'd think memory. test each stick alone or see how stable the system is with 1 stick at a time

[+Gary7 Subscriber²]

14 minutes ago, xendrome said:

yeah being NTOSKRNL it's likely a memory or voltage/power issue, I'd think memory. test each stick alone or see how stable the system is with 1 stick at a time

It must be quite common in the preview builds as That is what caused my last Blue Screen. I think it is the way that Win 10  is handling the memory and not a hardware issue.

[xendrome]

8 minutes ago, Gary7 said:

It must be quite common in the preview builds as That is what caused my last Blue Screen. I think it is the way that Win 10  is handling the memory and not a hardware issue.

He never said he was using a preview build. Infact he is on 10586 as noted in the BSOD dump reads

[+Gary7 Subscriber²]

Just now, xendrome said:

He never said he was using a preview build.

I should have seen that in his data, OK then this must be a common problem with Win 10 as I had it with the build he is using as well. I never had any types of memory problems with Win 7.

[Nagisan]

Going to run a memtest while at work tomorrow and Thursday with one stick each; should get a good 9 hour test for both of them and I'll report back with the issues. I don't think it's going to be the memory given the earlier test but you never know I guess.

[DevTech]

22 hours ago, Nagisan said:

Recently rebuilt my computer (replaced most parts of it) and everything seemingly worked fine.

What are the old parts and what are the new parts?

 

Memory test is a waste of time. If it's the memory, it would be a compatibility issue with the mobo. Also, depending on the type, you might need to either overvolt or undervolt it slightly.

 

1. Most likely it is a hardware compat issue.

 

2. Next likely is a device driver with kernel access which is anything on the PCIe/PCI bus.

 

3.A. Something is overheating such as CPU (bad cooler contact) or chipset or CPU power

 

3.B. Power glitches.

 

4. Firmware on SSD needs updating.

 

[techbeck]

1 hour ago, Nagisan said:

Going to run a memtest while at work tomorrow and Thursday with one stick each; should get a good 9 hour test for both of them and I'll report back with the issues. I don't think it's going to be the memory given the earlier test but you never know I guess.

Like mentioned before, if you have more than one stick of memory, pull all of them cept one, test.  If there is not BSOD, then try another single stick. Test again.  May help narrow down if there is a memory issue faster.

 

Could still be a driver issue as well.

[Nagisan]

Old

CPU - i7-970

MOBO - GIGABYTE GA-X58A-UD3R

RAM - G.SKILL Ripjaws X Series 24GB (6x4GB Tri-channel DDR3)

GPU - EVGA GTX 770

PSU - Corsair HX850

SSD - Samsung 850 Evo

 

New

CPU - i5-6600K

MOBO - MSI Z170A SLI Plus

RAM - Corsair Vengeance LPX 16GB (2x8GB Dual-channel DDR4)

GPU - EVGA GTX 770

PSU - Corsair HX850

SSD - Samsung 850 Evo

 

1. Everything is "compatible" but there may be some sort of conflict between two components

 

2. The only PCIe/PCI device I have is my video card, drivers are of course up to date

 

3. It could have been power issues due to the dynamic power settings of my motherboard. I have (since making this post) set the CPU Core Voltage to 1.3v which is both higher than default and not dynamic so I'm going to run like that for a bit and see if it helps. I've ran Prime95 on small FFTs for about an hour with a max temp of ~60c on the hottest core (after voltage changes)

 

4. Just installed Samsung Magician and my SSD has the latest firmware

 

Think I'm going to run the memtests anyway as there's no reason not to run it on each stick if I have nothing else to specifically check at the moment.

Edited July 20, 2016 by Nagisan

[Danielx64]

I know this may sound silly but is your memory listed on https://www.msi.com/Motherboard/support/Z170A-SLI-PLUS.html#support-mem? Also it may be worth checking if the voltage is set to auto and the motherboard is getting the correct voltage. Also I would have all the voltage settings to default or auto. Adding too much voltage can damage hardware.

[DevTech]

9 hours ago, Nagisan said:

Old

CPU - i7-970

MOBO - GIGABYTE GA-X58A-UD3R

RAM - G.SKILL Ripjaws X Series 24GB (6x4GB Tri-channel DDR3)

GPU - EVGA GTX 770

PSU - Corsair HX850

SSD - Samsung 850 Evo

 

New

CPU - i5-6600K

MOBO - MSI Z170A SLI Plus

RAM - Corsair Vengeance LPX 16GB (2x8GB Dual-channel DDR4)

GPU - EVGA GTX 770

PSU - Corsair HX850

SSD - Samsung 850 Evo

 

1. Everything is "compatible" but there may be some sort of conflict between two components

 

2. The only PCIe/PCI device I have is my video card, drivers are of course up to date

 

3. It could have been power issues due to the dynamic power settings of my motherboard. I have (since making this post) set the CPU Core Voltage to 1.3v which is both higher than default and not dynamic so I'm going to run like that for a bit and see if it helps. I've ran Prime95 on small FFTs for about an hour with a max temp of ~60c on the hottest core (after voltage changes)

 

4. Just installed Samsung Magician and my SSD has the latest firmware

 

Think I'm going to run the memtests anyway as there's no reason not to run it on each stick if I have nothing else to specifically check at the moment.

''1. Everything is "compatible" but there may be some sort of conflict between two components"

 

Yeah that.

 

"2. The only PCIe/PCI device I have is my video card, drivers are of course up to date"

 

You have tons of PCIe devices for which any driver can write into kernal memory and clobber the O/S.

 

I've done it way too many frustrating times myself!

 

- Intel - chipset driver + GPU driver + disk driver + USB 3 driver (all PCIe)

 

- LAN driver (prob PCIe0

 

- Audio driver (prob PCI)

 

Sometimes for some mobos it is useful to install IOBIT Driver Booster which can sometimes detect obscure drivers. After doing this you can just uninstall it to get it out of your way.

 

3. don't over volt the CPU more than 10% - check your Corsair specs to see if it has a voltage range skewed towards needing more voltage. Mobo BIOS typically report RAM voltage higher than actual.

 

4. good

 

5. Make sure you have a decent page file size on the Samsung so you can get a full dump. Make it 32 gigs during testing.

 

 

 

 

[Nagisan]

7 hours ago, Danielx64 said:

I know this may sound silly but is your memory listed on https://www.msi.com/Motherboard/support/Z170A-SLI-PLUS.html#support-mem? Also it may be worth checking if the voltage is set to auto and the motherboard is getting the correct voltage. Also I would have all the voltage settings to default or auto. Adding too much voltage can damage hardware.

Memory is not on the list but everything I've found says those lists are mostly useless these days, all it is, is memory that the mobo manufacturer can guarantee have zero issues with the motherboard but is not all inclusive due to the amount of different memory modules out there.

 

As for voltage, only reason I turned it up to where it i, is to possibly find out if lack of voltage (due to the automatic settings) was causing issues, was planning on lowering it after running for a week or so.

 

4 hours ago, DevTech said:

Sometimes for some mobos it is useful to install IOBIT Driver Booster which can sometimes detect obscure drivers. After doing this you can just uninstall it to get it out of your way.

I did try this software, it came up with about 8 updates to various devices (chipset and whatnot), my only concern is most people seem to agree that drivers from the motherboard manufacturer are the safer bet because they have tested those versions of the drivers on the motherboard or may have custom modifications applied to their drivers. Is that not the case, would it be better to install the device manufacturer drivers instead of the motherboard manufacturer drivers?

[DevTech]

On 7/20/2016 at 1:10 PM, Nagisan said:

Memory is not on the list but everything I've found says those lists are mostly useless these days, all it is, is memory that the mobo manufacturer can guarantee have zero issues with the motherboard but is not all inclusive due to the amount of different memory modules out there.

 

As for voltage, only reason I turned it up to where it i, is to possibly find out if lack of voltage (due to the automatic settings) was causing issues, was planning on lowering it after running for a week or so.

 

I did try this software, it came up with about 8 updates to various devices (chipset and whatnot), my only concern is most people seem to agree that drivers from the motherboard manufacturer are the safer bet because they have tested those versions of the drivers on the motherboard or may have custom modifications applied to their drivers. Is that not the case, would it be better to install the device manufacturer drivers instead of the motherboard manufacturer drivers?

It is not always practical, but when possible, get RAM from the Mobo Manufacturer Approved List. There are still odd incompatibilities in 2016 and for some reason this type of issue almost NEVER get fixed by a BIOS update. Often you can get a good idea which brand of RAM the mob manufacturer seems to prefer by scanning the list and noticing that 1/2 the list is one particular brand etc.

 

When there is no way to read the tea leaves, then G.Skill is the safest brand.

 

"most people seem to agree that drivers from the motherboard manufacturer are the safer bet"

 

Not even the slightest bit curious which street in Roswell, New Mexico contains "most people" but Alien Abduction problems have obviously damaged their bioware processing units.

 

Mobo manufacturers are notorious for repositories of dusty old drivers from a forgotten age. Good for Archeologists. Of interest to Indiana Jones when a certain version of driver for his Crystal Skull contains an Easter Egg showing the location of the Millennium Falcon.

 

[Nagisan]

1 hour ago, DevTech said:

Not even the slightest bit curious which street in Roswell, New Mexico contains "most people" but Alien Abduction problems have obviously damaged their bioware processing units.

 

Mobo manufacturers are notorious for repositories of dusty old drivers from a forgotten age. Good for Archeologists. Of interest to Indiana Jones when a certain version of driver for his Crystal Skull contains an Easter Egg showing the location of the Millennium Falcon.

"most people" referring to what I could find via google with most results leading to people recommending drivers from the Motherboard manufacturer rather than those from each device manufacturer. It's notoriously difficult to find answers to that question though given how most of the wording is pretty interchangeable.

 

Since my last post I have completed testing both ram sticks, 8 hour memtest on each with zero issues found so it's almost definitely NOT the ram itself. I also went ahead and updated all the drivers with the latest that the device manufacturers have available (Realtek for ethernet, Intel for chipset, etc). I have not had any BSODs since making this thread but as can be seen from the two crashes I had minidumps from they weren't exactly frequent.....here's to hoping one of the many things I've done have fixed the issue.

[DevTech]

4 hours ago, Nagisan said:

"most people" referring to what I could find via google with most results leading to people recommending drivers from the Motherboard manufacturer rather than those from each device manufacturer. It's notoriously difficult to find answers to that question though given how most of the wording is pretty interchangeable.

 

Since my last post I have completed testing both ram sticks, 8 hour memtest on each with zero issues found so it's almost definitely NOT the ram itself. I also went ahead and updated all the drivers with the latest that the device manufacturers have available (Realtek for ethernet, Intel for chipset, etc). I have not had any BSODs since making this thread but as can be seen from the two crashes I had minidumps from they weren't exactly frequent.....here's to hoping one of the many things I've done have fixed the issue.

Hope it works out.

 

Keep updating the drivers as they come out and eventually issues will tend to decrease. Windows 8, and hence Windows 10 added a lot more power management states on top of an already existing nightmare for device driver programmers. You can fill a wall with the chart of different power conditions for a Kernel Device Driver. Most device drivers use a whole bunch of shortcuts to lump together various state management transitions, otherwise they just could not afford to cost out the programming time. As long as your "E.T. Phone Home" in Windows 10 is turned on, the crash telemetry will eventually end up on a bug list in the office of some overworked guy somewhere in the world and if enough people trigger that unexpected state transition, the next driver update will have your patch. Hence it is like a giant world-wide self-repair loop.