Memory_Management BSOD on Clean boot, VERY specific circumstances

philc43 said:

Thanks for your explanations. Can I ask you to do some more testing of the Shutdown process?

Please have a look at the following tutorial and concentrate on options 8 and 9 which give you various ways of shutting down the computer.
Shut Down Computer in Windows 10

Can you determine if there is any type of shutdown which does not trigger the problem? For example compare shutdown /p with shutdown /s and with shutdown /s /f /t 0

I will give it a go, the only problem is that the BSOD only seem to happen when I switch my computer on first thing in the morning. If I turn it off now, it will boot perfectly. I've never successfully triggered a BSOD on boot during the day.

I won't know until tomorrow whether there is any difference.

It's one of the reason that I thought that this might be a paging file problem. I thought that maybe the PC would BSOD if the date that it shut down on was different form when it resumed, or that maybe something was trying to update itself the first time that it ran on any given day and would cause the BSOD.

If I turn it off now, it will boot perfectly. I've never successfully triggered a BSOD on boot during the day.

That statement makes me think you might have a motherboard or component failure which only shows up when the system has gone cold - i.e. it's a bad connection somewhere that only shows up when cold. Once the system is warm metal expands slightly and connections can start working again.

philc43 said:

That statement makes me think you might have a motherboard or component failure which only shows up when the system has gone cold - i.e. it's a bad connection somewhere that only shows up when cold. Once the system is warm metal expands slightly and connections can start working again.

I'd thought of that too, but I dismissed it for several of reasons.

1) It doesn't do it if there is only 1 stick of RAM in the system, only 2.
2) It didn't do this under Windows 7
3) I left it switched off for 3 or 4 hours on the same day and it didn't BSOD. Which should be enough time for it to cool down.

Tracking down these problems is never straightforward but I admire your logic and tend to agree, especially reason 2. However, W10 is probably stressing your system more than W7 did.

I keep trying to think of things related to the two versus one memory sticks - when they are running in dual channel mode maybe that is what is stressing the system and causing the BSOD. I had dismissed this at first because you had tried slots where they would not run in dual channel mode but perhaps there is something incorrectly set in the BIOS for the memory timings. Have you been through these settings and looked at whether you have XMP (extended memory profile) activated?

philc43 said:

Tracking down these problems is never straightforward but I admire your logic and tend to agree, especially reason 2. However, W10 is probably stressing your system more than W7 did.

I keep trying to think of things related to the two versus one memory sticks - when they are running in dual channel mode maybe that is what is stressing the system and causing the BSOD. I had dismissed this at first because you had tried slots where they would not run in dual channel mode but perhaps there is something incorrectly set in the BIOS for the memory timings. Have you been through these settings and looked at whether you have XMP (extended memory profile) activated?

I'm not sure what XMP is.

I've tried shutting down using shutdown /s /f /t 0 a few times but I've not been able to establish if this as made a difference or not.

I used Autoruns to take disable a few processes\services\programs, and at the time when I normally get the BSOD my screen went nuts.

It repeatedly flickered between the whirling dots and a plain black screen. I shut down and rebooted and logged in as normal but got a Critical Service Failed BSOD about 20 seconds after logging in. I then rebooted and all of the icons on my start menu were missing (No Apps, no tiles). So I did a system restore. Which has put my settings back to where they were some time last week.

This was likely my own fault as I probably took out something important (I think that it was the smart card login service). So I'm not going to know if the shutdown code is effective until tomorrow.

philc43 said:

That statement makes me think you might have a motherboard or component failure which only shows up when the system has gone cold - i.e. it's a bad connection somewhere that only shows up when cold. Once the system is warm metal expands slightly and connections can start working again.

I can check this by booting to the BIOS, and leaving it there for a while. That will warm the PC up. I'd like to check other things first though.

My PC has quite a high specification cooling system, so the difference between a cold start and idling on the login menu isn't very much. It's not much over room temperature. There are 2 fans drawing air in from outside and blowing it across the RAM, so the sticks don't get very hot even during gaming.

I used to do a lot of 3D rendering, so I'd stress out my system components with continuous calculations, so when I brought this PC I made sure that the cooling was over-specced for the load that I was going to put on.

OK, so I'm not really getting anywhere with this problem.

I downloaded Autoruns and used it to disable some programs and services to stop them from loading at startup to see if this made any difference, and that destabliized things a little. I stopped getting the BSOD at startup started getting various different BSOD a few minutes after I'd logged in. Some times I'd get one if I logged in and just left the PC without running anything manually. Different ones each day, and then I lost everything on my start bar. I restored to a previously saved configuration from a couple of weeks ago (Befor eI made many changes) and continued to try different options in Autoruns. Now the BSOD on the clean boot is back, BUT it's not happening every day.

My wireless keyboard also started lagging\not responding. It's back to normal now.

I think that it's definitely a program\driver or service that's causing this, rather than a hardware problem for the simple reason that taking things out of the startup settings changes the order\time at which other program\driver or services run. Which is causing the BSOD to happen earlier or later in the boot process.

When I restored the settings back to how they were a week ago (Thus restoring the boot order for program\driver or services) the BSOD went back to the EXACT same place as before.

There's no reasonable way that a bad PSU, for example, would be so specific as to fall over at the exact same moment in the boot process every day for a month (Right down to the 1\2 second), and then to change when the services at startup changed, and then to change back when the programs at startup were restored.

It's also definitely not my anti-virus as when I did a fresh install a few months back I didn't install the anti-virus until a day later, and I had a BSOD on my first clean boot after being shut down over night.

Though it could be Windows Defender or something that Microsoft puts in?

I have gone back through this thread to see if I had missed anything. I keep coming back to the memory. Have you ever run Memtest86 for the full 8 passes? With your 16GB it will take all day to complete the tests and reach the minimum of 8 passes but I think it is worth doing (to rule out any possibility of errors in the RAM). Any sign of even a single error and you can stop. You could test them one at a time overnight which might be easier to do. You should really test them in each slot too to rule out any problems with the MB slots so it would be several nights worth of testing.

Another question: Have you ever tested in safe mode?

philc43 said:

I have gone back through this thread to see if I had missed anything. I keep coming back to the memory. Have you ever run Memtest86 for the full 8 passes? With your 16GB it will take all day to complete the tests and reach the minimum of 8 passes but I think it is worth doing (to rule out any possibility of errors in the RAM). Any sign of even a single error and you can stop. You could test them one at a time overnight which might be easier to do. You should really test them in each slot too to rule out any problems with the MB slots so it would be several nights worth of testing.

Another question: Have you ever tested in safe mode?

I ran MemTest with each stick individually for about 5 hours, and with both together for about 5 hours. I don't remember exactly how many passes that was, but things are a little complicated and I can't run the PC over night.

The PC only seems to BSOD after a clean shutdown and after it has been left switched off for quite a while, and not 100% of the time. I've tried booting it to safe mode and it hasn't BSOD, but I really can't be sure if this is significant or not.

I've never successfully triggered a BSOD without leaving the PC switched off over night, so I can only really try one thing per day, so it's slow going.

The fact that it only seems to BSOD IF it's been left switched of for a long time tells me that there is a software process involved that checks the time, or records a log, or tries to check itself for updates online. Something that happens on a scheduled basis.

Also, because the BSOD changed when I changed what loaded on startup tells me that this is not a hardware problem.

I've tried booting it to safe mode and it hasn't BSOD, but I really can't be sure if this is significant or not.

I think this is significant especially if has been done after an overnight shutdown and helps us to rule out hardware. The clean boot process is going to be your best troubleshooting tool, the difficulty is doing it when you know a BSOD is triggered.

Please review the process here: https://www.tenforums.com/tutorials/4...conflicts.html and if it still happens with all the startup items removed and all the non-Microsoft services disabled we know it is a Microsoft service that is causing the problem and we can start disabling those a few at a time. If it does not happen then you can start adding back services and then eventually the startup items until it does happen.