Constant crashing since upgrading to Windows 10

[CelticBorealis]

Hi everyone,

So, the issue is that I upgraded to Windows 10 in late September and since then I've had constant problems with crashing.

I originally thought that the problem was caused by Steam. (I posted a topic on this here.) This was because most crashes seemed to be happening while playing games.

I'm now 100% certain that it isn't anything to do with Steam because my computer crashes when playing non-Steam games (and I'm still playing ancient games like Age of Empires II, so it's nothing to do with my system specs) and when I'm not playing any games at all. Right now, it's still crashing and I haven't got Steam or any games installed as I have just done a fresh install of Windows 10. The frequency of crashes, however, is increased while playing games.

Nature of crashes

Most crashes are a freeze. Everything will freeze in place. If I'm running anything that produces sound I'll get a sort of 'brrrrrr' sound coming from the speakers, forcing me to press the reset button. Sometimes it just turns off completely. I've never had a blue screen of death.

This has only started happening since upgrading to Windows 10 - I had no problems with Windows 7. It's been going on for nearly three months now. I've been keeping a detailed log of every crash and what I was doing when it crashed along with times.

There is little to no consistency as to when it crashes - I can be running multiple programmes or I can be doing nothing at all. It'll still crash. The frequency is increased if I am playing games, certainly, but sometimes it runs games absolutely no problem. Some days I have no crashes and it's perfect; other days it will crash (literally) every 20 minutes.

Sometimes it will do a sort of mini-crash, where it will freeze (and make the brrrr sound) for a couple of seconds and then resume and be fine again.

What I've done

I've followed all the advice I've been given here and from people IRL. Among other things:

• I've updated everything that can be updated.
• I've run the memory diagnostics tools and I'm getting no reported problems.
• I've run the system file checker and I'm getting no reported problems.
• I've taken it to a repair shop and got given a £95 bill for them to tell me there were no problems. They said they left the machine streaming the internet overnight and it was still happily running the next morning.
• I've purchased a new (more powerful) PSU because I thought it may be that I was on the edge of what my old one could manage.
• I've purchased a new hard drive and I did a clean install of Windows 10 on it. My old hard drive (which I think is working fine) is currently in a box, but I would like to put it back into my computer at some point. I was planning on getting a new hard drive anyway.
• I've thoroughly cleaned my computer and put fresh thermal paste on the CPU.

Overall I've spent just over £200 trying to sort this problem out. It's not really money I have. I'm self-employed and rely entirely on my computer for work. Without a computer I don't get paid, and I can't buy new stuff without money…you see the problem here.

My system specs

My system specs are:

• AMD FX6300 6-core CPU
• Sapphire Dual-X Ra270x GPU
• 2 x 4 GB RAM
• MSI 970A-646 motherboard

Possible problems

One observation it that it seems to be running quite hot at the moment. I don't know why - it's very clean, and it's winter so my office is quite cool at the moment. I say it's running hot - I've got the CAM monitor software installed and it's saying my CPU is running between 35 and 40 degrees Celsius (all I've got open is Google Chrome at the moment) and my GPU is at 30. However, the fans (I think the CPU fan) seems to be spinning very fast and making quite a lot of noise - much more noise than I would expect it to make while I'm doing non-intensive tasks.

Could there be a problem with the ISO disc I installed Windows 10 from? Could I have installed a flawed operating system? I burnt the DVD on a friend's computer.

Could there be a problem with my CPU or motherboard? The motherboard doesn't have integrated graphics, so I'm fairly certain my GPU is perfectly functional. I've had motherboard problems before. I first built the computer in summer 2014 and the motherboard I received was faulty, so I got a new one on warranty. It began playing up again earlier this year (annoyingly just outside the warranty period). I brought yet another one in January 2016 (exactly the same model) and I've had no problems with it. Once again - all these problems have started since upgrading to Windows 10, so I'm unsure as to why my hardware would start playing up.


Sorry this is such a long post - I'm utterly confused by this problem and it is driving me mad. It's costing me a lot of money and many, many hours of my life when I need to be working.

If anyone has any suggestions I'm all ears and basically willing to try anything.

Many thanks for taking the time to read this.

CB

 

[MeanMachine41]

It can be quite a long troubleshoot CelticBorealis and time is not what you have if its effecting your business. If you have the time then a series of stress tests on your system may reveal the culprit and I can help with that.
My advice is to revert back to W7 if you had no issues, and when you have the time then sort out why W10 is playing up. You are not alone with W10 issues.

 

[SkyNetRising]

If windows 10 is such a pain for you, why don't you go back to windows 7 ?

Anyway - I'd look in to Event viewer and check for errors/warnings.

 

[Shaun o]

Well I think the problem is to do with how windows 10 set it`s self up Cb.

I don`t know if anyone has ever mentioned to you about Irq`s.

Interrupt request ques.

So i will explain how it works for you to get an understanding of how it can effect your system, but also cause the freezing of your system as it stands probably at random times when in a windows session.


In basic terms in order for the cpu to run multiple drivers or programs at the same time or hardware devices connected to the system via all of the interfaces and ports provided via the motherboard.

It does not focus on running one task or driver ect.
Instead via the cpu of the system the cpu it`s self rapidly switches between running tasks, hardware or drivers running on the system.

Obviously windows along with the motherboard has to keep track of each driver and hardware device running.

So it uses an IRQ number set for each device, driver or bit of hardware connected to the motherboard via the interfaces provided.

The cpu basically controls the timing, when a driver requests for a hardware device an action, or operation to be performed.
So it will send data or a signal to the cpu based on the IRQ number flagged or fixed to use the device.
But also to receive the data or result from it when the request or operation is performed.
Storing the result or action to system memory or a storage device.

Each hardware device will have it`s own fixed IRQ number or assignment, or should do if windows does it`s job of setting a unique Irq number for each hardware device.

In basic think of it like you phoning a good friend, but someone has already dialed his number and is talking to him on the phone.

You get a busy signal, saying no connection can be made to the number you dialed.
This is what happens when for example windows selects the same designated IRQ set number for two or more hardware devices.

The result being the system may crash or freeze. CB.
Or in windows you may be presented with a error message of the hardware, or it`s driver failing to respond to a request or instruction asked of it.

So how do you fix the problem.?
Basically an IRQ is a resource of the motherboard.
Remember each bit of hardware or an interface on the motherboard each requires a set IRQ number or assignment.

There for in the bios of the motherboard you simply disable interfaces of the motherboard using resources or Irq numbers that you do not need or use.

Examples can be, Lpt 1 and serial ports assigned an Irq in the bios.
Reducing the amount of Usb ports of the motherboard that are enabled.
Individual Sata ports of the motherboard that do not have a Drive connected to them set to disabled mode.

Any conflict between two hardware devices based on the Irq number set if the same Irq number is selected can cause as said a reason as to thy a system may freeze randomly.

In fact the sound being stuck in a loop and stuttering is a good indication it`s self that your problem is with more than one hardware device on your system wishing to, or fighting for the same IRQ number to be allocated to it via windows.

You can look in device manager of windows and under properties It will list what Irq number each device is set to via windows os.
All you have to do is find three or more devices set to use the same Irq number.
And that the cause of the system freezing.

For example say the Pci-e graphics card in the system was set to Irq 9
But also the sound chip of the motherboard for audio was also set to Irq 9.
And also say some of the Usb ports of the system, motherboard.

It explains why the system is freezing and you are left with the sound looping like a machine gun is being fired or stutters in a loop.

New motherboards are more capable of managing and setting up hardware devices and assigning Irq numbers far better than in the old days for the resources of the motherboard.

But even so between windows Os being installed and the motherboard there are exceptions where it can still get resource settings wrong, and as said be the cause to as why a system may lock up, or freeze at random times or become unstable CB.

 

[CelticBorealis]

Thank you for your replies - greatly appreciated.

Unfortunately I can't go back to Windows 7. The reason being is that my copy of Windows 7 was licensed through my old workplace, but I no longer work there. I got given the disc from work along with the activation code. There was quite a long gap between me building the computer and installing the operating system because I was working for the same company abroad for several months. In that gap I managed to lose the activation code, and I left that job shortly afterwards (so I couldn't really ask for the IT guys to give it to me again). Up until recently I've been using a perfectly legal copy of Windows 7, only it was unregistered which meant I couldn't upgrade it normally via Windows' promotion. Windows 10 I purchased myself, and I'm stuck with it now.

I forgot the mention the error messages. According to the Event Viewer every single crash is (exclusively) listed as: Kernel-Power (Event ID 41, Task Category 63).

I've googled this, and it seems to simply be a generic error message Windows creates when it doesn't know what has caused an unexpected turn off. Clearly (judging by what I've read online) this isn't a problem unique to me and can be caused by a huge variety of problems. I'd tried various solutions I've read online about dealing with these already. I can try them again, but I see no reason to think that they would fix things this time.

Shaun o - thank you for your detailed suggestion. I didn't know about this. I'll try this in the evening and let you know if this fixes things.

I'll look into stress testing too. That will probably have to wait until the weekend.

 

[SkyNetRising]

So going back to windows 7 is not an option.
It's possible that upgrade process to windows 10 didn't go as supposed.
You can try doing clean installation of windows 10.

BTW - Have you checked your PC against viruses/malware?

 

[CelticBorealis]

I've done three clean installs since September. The latest clean install was done just a few days ago. Even with no additional software installed it was still crashing. The latest clean install was done on a brand-new hard drive - my old hard drive is sat in a box at the moment, and I plan to reinstall it at a later date. I can try burning a new ISO DVD and doing a clean install from that, though I've burnt the DVD twice already.

I've not done any virus checks since upgrading from Windows 7 (and I haven't got any virus checking software installed yet). I used to virus/malware/etc. check regularly, but given that I've been clean-installing Windows 10 I shouldn't have any. The only reason I could have taken on a virus is if the install disc came with one, but I downloaded the ISO straight from Microsoft so I'm guessing that's fine.

 

[CelticBorealis]

Following on from Shaun O's answer I tried exploring the priority queue properties and couldn't see anything amiss.

Nobody seems to think it's a software issue, and that all these crashes just happen to be coincidental with upgrading to Windows 10.

I'm thinking that given that I've replaced my PSU and HDD (though I will be reinstalling my old HDD once I've got the problem fixed) that I should replace my motherboard. It's still under warranty. I'm going to buy a new one today, so we'll see how that goes...

 

[CelticBorealis]

New motherboard arrived today. I installed it first thing.

Everything seemed to be running better, but then, after a few hours, it crashed - just switched itself off. This is a different sort of crash (before it would freeze and force me to press the reset button). Now it's just turning itself off completely - no BSOD and no error messages of any kind.

In the event viewer I'm getting the kernel_power error, which is exactly what I had before.

I'm at a complete loss. I've replaced the PSU, the HDD and the motherboard now. None of them have achieved anything. I've done every bit of diagnosis everyone has suggested and it's revealing nothing. I've taken it to a repair shop and they don't think anything is wrong.

Any further ideas at all?

Any suggestions will be greatly appreciated.

 

[SkyNetRising]

What's the model name of PSU, you're using?
Have you done any OC?

 

[CelticBorealis]

My PSU is a Corsair CX750, which is way more power than I actually need. I was using a 500 W PSU (I think an MSI one - it's in a box in my cupboard now). This was perfectly adequate for my setup and wasn't causing me any problems.

I haven't done any overclocking. My new motherboard has various in-built overclocking features (as did my old one), but I've not explored them at all. I don't do anything that would require it.

I know it's not a GPU problem because I swapped my GPU into a friend's computer and he experienced no problems with it. The only hardware left is the RAM and CPU, but I can't think of anything I've done that would stop them working properly. Visually they look absolutely fine. All of this has happened since installing Windows 10. I've done three clean installs and checked all the updates.

 

[MeanMachine41]

The kernel_power error is reported because the system was not shut down correctly. Do not concern yourself with this.

As the system is shutting down for no apparent reason, no BSOD then I think you may have instability issues with your system and its time to run some diagnostic tests with a Stress Tester.

Download AIDA64 and HWMonitor and put them side by side on your desktop.
In AIDA64 go to the Tools menu and launch the stress tester. In the check boxes, run test for CPU,FPU, and Cache. Other tests will be run separately later.
Run the test for 10mins initially and stop the test if temps go over 80C.
Observe your temperatures in HWMonitor and take screen shots at the 10min mark.
Upload the screen shots to IMIGUR or your favorite file host.
In IMIGUR go to Images and obtain the BBurl for a link here to analyse the results.

I will want to check all rail voltages in HWMonitor with 100% Utilization on all cores and core temperatures during the test.

You can also do independent testing on your RAM using Memtest86+ by booting from a USB stick.
If you do then test all RAM together and independently for at least 3 passes.

 

[CelticBorealis]

I greatly appreciate this. I've been concerned about temperatures, which I think are much, much higher than they should be.

I did what you said. I actually did it twice, because this is bizarre.

First Test

Temperature - CPU temps shot up to 78C and stayed there the whole time. HWMonitor said they went up to 84C at some point, but it must have been extremely brief as I was watching the screen the whole time.


Powers


Voltages


Second Test

Temperature - This was noticeably different. It was substantially quieter, and the temperatures didn't go much above 40C. I have noticed the computer running rather hot even while doing non-intensive tasks. This is much cooler


Powers


Voltages


The reason I did a second test is that my computer crashed shortly after I did the first one. However, I noticed upon restarting that my system was working well. The computer was quiet, and the temperature readings were low (not much more than mid-20s/30s. The computer remained this way for the whole day. In fact, I deliberately began running intensive applications. I do a fair amount of data processing for work, and so I started working with some huge data sets. I tried doing some video editing and Photoshop - the computer handled all this without too much problem at all. I had no crashes or issues at all. In played a few games in the evening, and again - no problems. Temperatures never crept above 50C, and most of the time remained below 40C.

This is how it should be, and how it was before.

Upon switching it back on this morning it's reverted back to crashing. I crashed just a few minutes after switching it on. CPU temperatures shot up to the 60s/70s immediately after start up, and occasionally breached 80, even while the computer was doing next to nothing.

I realise I should have taken the screenshots while the test was still running rather than stop it and still take them. I'm going to run it again today and do it properly this time.

 

[MeanMachine41]

I'm suspecting the PSU may have something to do with this. Voltage output on all rails should be +/- 5% and I notice your overvolting on the 5V and 12V rail.

This could be a reason the system shuts down and why temps start to rise.
I can't be 100% sure but a failing PSU can cause this. You should swap it out with a known good working unit of the same or higher Wattage if you can.

What is your PSU Age,manufacturer, Wattage, and tier rating.?

Never mind I know and I see you have another in your cupboard. Try that.
The Corsair CX750 is bronze rated and have poor 2nd tier capacitors and load regulation issues.
If I were you I would upgrade your PSU to a Seasonic gold rated unit of the same Wattage.
Best on the market and worth the investment.

 

[CelticBorealis]

Thanks for looking at this.

I thought it might be a PSU issue too, though I hadn't realised about the voltages. The old (500 W) PSU was salvaged from a friend's computer, and worked without problems for several years. The 750 W unit that's in now was purchased only a couple of weeks ago (I have an extraordinarily limited selection in my local shop, and I wasn't willing to wait for an Amazon delivery). I've had exactly the same issue pre- and post-PSU change, so I suspect my old PSU is working just fine still. If it is the PSU then I've been extraordinarily unlucky.

I've replaced so much of the computer now, and I'm basically out of money. I've spent approaching £300 on new components (motherboard, PSU, HDD) and a wasted trip to the repair shop (which achieved nothing) trying to fix this problem. That, and the lead up to Christmas, has rather hurt the old bank balance. It's getting to the point now where building a new computer from scratch would have been cheaper.

It's just so weird that yesterday the machine worked perfectly without me having done anything. Today I had just the one crash (first thing this morning), but I can see that it's running pretty hot now and all I'm doing is web browsing. I've been out most of the day, so the computer hasn't been used much so I can't really ascertain if it's working properly...

 

[MeanMachine41]

You shouldn't be out of pocket CelticBorealis. Get an RMA from Corsair and return the unit from the place of purchase.

The PSU is about the most important part of your system, Its the life blood and can also cause damage to other hardware if it fails or is overvolting.
Bronze rated PSUs are generally budget buyers choices for general purpose use or thrown into pre built systems. Gold or higher tier units are better quality and more efficient.
Read the PSU fact sheet in my signature.!!!

 

[CelticBorealis]

I really, really appreciate your help on this.

I think I'll wait until after Christmas and see what happens in the January sales. Unfortunately work tends to dry up for me heading towards Christmas because I am self employed. I tend have an extremely busy November as my clients like to get everything out of the way before their Christmas break. Unfortunately, that tends to mean December is quiet.

The article in your signature was interesting. It does help to explain why I have never had to heat my office, even in the middle of winter! This was the first (and to date only) computer I've built myself, so inexperience probably didn't help.

I have possibly discovered the main problem, however, and that is the RAM. It is really, really strange.

What I did was remove a stick of RAM to find out if one of them happened to be broken. Upon starting up again it said 'memory changed' and then gave me the option of setting it up manually or using defaults.

I selected defaults. The computer then operated perfectly - running cool and quiet with no crashes.

When I shut down and started up the next day it was back to running extremely hot, loud and crashing regularly. I shut down, reinserted the RAM (so back up to 2 x 4 GB RAM), selected defaults again. Once again, the computer ran just fine...

...until the next day, when I started up and it was back to crashing. I did the same thing - removed a stick of RAM and selected defaults. All good again.

I've done this several times now. It doesn't seem anything is wrong with either stick of RAM, but it does seem like every time it shuts down something changes in the settings (somewhere) to make it stop working again. Clearly selecting the default settings whenever I change the memory (it makes no difference whether I have two sticks or one stick of RAM in) fixes the issue.

Do you happen to have any idea what causes this?