Computer crashes, BSOD, freezes

[whysarx]

Hello.

I've been having this problem for some years now and i still havent managed to solve it. What happens is the computer randomly crashes, BSOD or just freezes, could be when gaming or just browsing. The computer is a bit old with the specs as follows:

AMD 925
HD 5850
Asus M4A785TD-V EVO
6GB @ 1600mhz

Very rarely do i get a dump file from the crashing/bsod but I have 3 which all point to ntoskrnl.exe.

I have tried doing a ram check with memtest86 and windows memory utility and they both came out okay. I tested the disk and it also appeared fine. I ran Prime95 in blend and still no crashes. I can't figure out whats crashing the system and am out of ideas.

Any tips?

 

[Shektron]

Drivers. Try re-installing the drivers - GPU drivers, especially.

 

[whysarx]

thank you for replying. I had already tried that with the two still available drivers and still crashes. I can make the pc crash after running furmark cpu burner (not the GPU part). ill test with prime95 with other options than blend.

Could it be a failing CPU? isnt that impossible or super rare?

 

[Shektron]

Did you monitor CPU temps? Try replacing the thermal paste and re-seating the cooler a bit more carefully, in case the temps are abnormal.

As for failing CPU, depends on how much voltage you run it at, the temps, and how long you've been using it at that.

 

[whysarx]

Did you monitor CPU temps?

Im running prime95 with the option that generates maximum heat but cpu doesnt pass 37ºC so far. will keep it testing to see if it rises or crashes the pc. I have very good cooling in this pc.

Try replacing the thermal paste and re-seating the cooler a bit more carefully, in case the temps are abnormal.


I replaced the paste 4 months ago and temps are okay.

As for failing CPU, depends on how much voltage you run it at, the temps, and how long you've been using it at that.

Im running it at stick with no overclock. The pc has 7 years. According to HWMonitor the voltages are as follow:

I find it weird that the furmark cpu burner (not GPU) makes my system crash but not prime95. Will keep it running for a bit more.

 

[Shektron]

Temps are pretty good, so it really is difficult tell, at least within my knowledge. Try to run other kinds of stress tests - both GPU and CPU. Run AIDA64, Unigine, PCMark, and other stress tests as well. Keep an eye on the GPU as well, and did you OC the GPU?

Considering current information, it really looks like a dying part, because everything else looks pretty normal to me.

 

[whysarx]

Still no crashes with prime95(30minutes). Temperature didnt pass 37c. Ill try using the other programs you suggested.

 

[johnbl]

in general for bugchecks: you want to update the bios and motherboard drivers from the motherboard vendors website.


for bugchecks that do not produce a memory dump, you would suspect the storage system
(and update the bios and motherboard drivers) but sometimes you also have to update firmware of a solid state drive and make sure there is plenty of free space on the drive. run crystaldiskinfo.exe to read the smart data of the drive. You should also consider putting the drives data cable on a different port and controller if your system has two SATA controllers. Generally, the slower one directly supported by the cpu chipset will have fewer bugs in the driver.


with freesing/lockups they can happen when a graphic driver gets messed up. The most common cause of this is overclocking of the grahics driver and conflicts between the graphics drivers sound system and another sound controller in your system. very common for the motherboard sound driver to cause this. most motherboard vendors did a audio driver update after windows 10 shipped.


there are other causes of the lockups but these are the most common. if these attempt at a fix don't work then you have to google how to force a memory dump via keyboard, make the registry settings and attempt to force a memory dump while the system is locked up. Generally you would want to run verifier and set debug flags and change the memory dump to kernel so the debug data is not stripped out of the memory dump.

you would also have to put the memory dump on a server and post a link so it could be read with the windows debugger.

 

[whysarx]

in general for bugchecks: you want to update the bios and motherboard drivers from the motherboard vendors website.


Im running the most up to date bios. The PC is quite old so there wont be anything new.

for bugchecks that do not produce a memory dump, you would suspect the storage system
(and update the bios and motherboard drivers) but sometimes you also have to update firmware of a solid state drive and make sure there is plenty of free space on the drive. run crystaldiskinfo.exe to read the smart data of the drive. You should also consider putting the drives data cable on a different port and controller if your system has two SATA controllers. Generally, the slower one directly supported by the cpu chipset will have fewer bugs in the driver.

Im not running an SSD. I am using a Samsung Spinpoint F3 500GB. Here's the crystaldiskinfo results:

with freesing/lockups they can happen when a graphic driver gets messed up. The most common cause of this is overclocking of the grahics driver and conflicts between the graphics drivers sound system and another sound controller in your system. very common for the motherboard sound driver to cause this. most motherboard vendors did a audio driver update after windows 10 shipped.

Im not running a overclock in this system and am using a Xonar DGX card with UnI drivers but also tested with official and same thing.


there are other causes of the lockups but these are the most common. if these attempt at a fix don't work then you have to google how to force a memory dump via keyboard, make the registry settings and attempt to force a memory dump while the system is locked up. Generally you would want to run verifier and set debug flags and change the memory dump to kernel so the debug data is not stripped out of the memory dump.

Im sorry i didnt explain it well. I meant like 80% of crashes dont generate a dump, but i still got 3 .dmp files as you can see in this picture.

===================

I have been running AIDA64 and still no crashes for 20min.

Another thing is the crashes are related to the type of work im doing. I noticed that I never or very rarely get a crash when playing dark souls 3 but when i play rocket league its very common.

I also did sfc /scannow and a dism repair image command.

 

[johnbl]

you clearly have a memory corruption problem shown in your bugchecks.
something has continued to use kernel memory pool after it was released back into the system.
your bugcheck codes indicate bad memory addresses being used.
memory address 20 would not be a proper memory address but is more likely a pool header size.

you should change the memory dump type to kernel, then run powershell or cmd and run
verifier.exe /standard /all
reboot the system and see if you can get a bugcheck. verifier will look for bad drivers and force a bugcheck if it finds one.
the data will be stored in c:\window\memory.dmp file

even old computers should have updated drivers even if you have to get them from the chipset vendor.
(thinking of the audio drivers. but if you can not find a update then disable any unused audio source to prevent memory corruption)

in general for bugchecks: you want to update the bios and motherboard drivers from the motherboard vendors website.


Im running the most up to date bios. The PC is quite old so there wont be anything new.

for bugchecks that do not produce a memory dump, you would suspect the storage system
(and update the bios and motherboard drivers) but sometimes you also have to update firmware of a solid state drive and make sure there is plenty of free space on the drive. run crystaldiskinfo.exe to read the smart data of the drive. You should also consider putting the drives data cable on a different port and controller if your system has two SATA controllers. Generally, the slower one directly supported by the cpu chipset will have fewer bugs in the driver.

Im not running an SSD. I am using a Samsung Spinpoint F3 500GB. Here's the crystaldiskinfo results:

with freesing/lockups they can happen when a graphic driver gets messed up. The most common cause of this is overclocking of the grahics driver and conflicts between the graphics drivers sound system and another sound controller in your system. very common for the motherboard sound driver to cause this. most motherboard vendors did a audio driver update after windows 10 shipped.

Im not running a overclock in this system and am using a Xonar DGX card with UnI drivers but also tested with official and same thing.


there are other causes of the lockups but these are the most common. if these attempt at a fix don't work then you have to google how to force a memory dump via keyboard, make the registry settings and attempt to force a memory dump while the system is locked up. Generally you would want to run verifier and set debug flags and change the memory dump to kernel so the debug data is not stripped out of the memory dump.

Im sorry i didnt explain it well. I meant like 80% of crashes dont generate a dump, but i still got 3 .dmp files as you can see in this picture.

===================

I have been running AIDA64 and still no crashes for 20min.

Another thing is the crashes are related to the type of work im doing. I noticed that I never or very rarely get a crash when playing dark souls 3 but when i play rocket league its very common.

I also did sfc /scannow and a dism repair image command.

 

[whysarx]

[strike]Im on my phone right now. The command you told me to do caused the system to bsod at startup always. How can i disable it? I tried verifieer.exe /reset but it said nothing was changed.[/strike]
Fixed

So I think youre right about the cause of the problem. The memory.dmp file is 800Mb in size, how should i analyze it to find the cause?

 

[whysarx]

Bluescreenview with minidump gives me this:

Is it really the mouse drivers crashing my system? If so how do I fix that? i use the drivers for macros and stuff and would rather not lose it but if theres no other solution then I will uninstall it.

 

[johnbl]

you would have to put the memory dump on a server like Microsoft onedrive, share the file for public access and post a link.

yes, certain mouse drivers can cause this problem under certain conditions. (usb mouse drivers can corrupt data on the usb bus, usb bus can be shared with the PCI/e bus which has your graphic card on it.)

your memory dumps show parameter 1 (the error code) to be 0xc0000005 which is generally a bad memory address being passed to a driver.

you have one error code 0xc000000e
which is A device which does not exist was specified.

since it was a kernel in page error it could mean that your drive is connecting and disconnecting over and over.
or has become unresponsive. normally you would check cable connections, update sata drivers or move the drive to a different sata port and controller. also check for firmware updates to the drive and make sure there is plenty of free space. if the drive firmware is behind in its garbage collection routines, you can boot into bios and leave it powered but unused. After 5 minutes, the firmware cleanup will run and try to clean up the drive.

note the timestamp string on the Logitech looks like a ok version.
the human interface drive has a bogus timestamp (in 1991)
current versions windows have some in process memory compression and tend to mess up the compression where the drive dates become corrupted. It also messes up the debugger and makes it hard to figure out problems

Bluescreenview with minidump gives me this:

Is it really the mouse drivers crashing my system? If so how do I fix that? i use the drivers for macros and stuff and would rather not lose it but if theres no other solution then I will uninstall it.

 

[whysarx]

Thank you very much for your answer. I removed the .sys files that were giving me issue and re-ran driver verifier and it passed!

However i think the pc is a lot slower, is it normal or placebo? Can i fix it?

Edit: Removed the sys files and disabled the device in device manager.
Edit2: Onedrive link with dumps: http://

 

[johnbl]

here is a problem: you last bios for this machine was in 2010, the specifications for USB were changed in 2012 and windows 10 will be using the new specs. The new specs require a BIOS update after about april of 2012 and your motherboard is out of support. basically windows 10 will run until it hits a bug in the USB or SATA then you will have to reboot. You might be able to disable your onboard USB ports and put in a cheap USB 3 or USB 2 card.
or go back to a older version of windows that you can get proper drivers for.
I guess you could go back to a PS/2 keyboard or reboot your machine every few hours and never let it sleep.

you bios info:
BIOS Version 2103
BIOS Release Date 06/18/2010
Product M4A785TD-V EVO
Version Rev X.0x


-------
you have a USB device that failed, looks like it was removed but its file handles were not closed so it never finished the remove.
InstancePath is "USB\VID_0000&PID_0002\5&33cee4ba&0&2"

some mice and keyboards have firmware updates that you have to apply.



first time I have seen this driver installed:
\SystemRoot\system32\drivers\tsusbhub.sys

I would also suspect this driver:
Asus ATK0110 ACPI Utility (l would look for a update)
\SystemRoot\system32\DRIVERS\ASACPI.sys Thu Nov 1 18:54:34 2012

depends on what the cause of the problem was. Sometimes usb problems leave drivers that are old, some of the old drivers respond to all events on the USB bus rather than just events for that driver. generally, these will just slow your system down and will not cause a bugcheck unless some timeout value is reached.
generally, I can see these by looking at a kernel memory dump, there are internal logs that track errors.
I have seen USB logs record 26 million event logs in 10 or 20 minutes.

you might be able to run usbview.exe and see if you have a problem with a usb subsystem. its source come with the windows device driver kit but people have made versions you can download. I use this one most of the time:
http://www.uwe-sieber.de/usbtreeview_e.html

most of the time, I just look at the kernel memory dump using the debugger and dump the various internal logs to look for a problem. I also look at the various file dates and BIOS dates.

also, if windows finds a bugcheck on your system it sends it to a Microsoft system for debugging and tries to push a fix back to your system or do additional debugging. you can find these special memory dump if you search your hard drive for files with a .dmp extension. for most machines, Microsoft might find the cause of the problem but be unable to install a fix (for various reasons, like special drivers required or BIOS update required) On my old machine the windows update would turn off sleep because of known bugs in the BIOS. My wifes asus machine has known bugs in the USB support chips and you just have to reboot the machine to fix the USB ports.

Thank you very much for your answer. I removed the .sys files that were giving me issue and re-ran driver verifier and it passed!

However i think the pc is a lot slower, is it normal or placebo? Can i fix it?

Edit: Removed the sys files and disabled the device in device manager.
Edit2: Onedrive link with dumps: http://

 

[whysarx]

Heres what the program says:

I think everything is fine though. I am just seeing a lot of ram usage going no where and i think thats what causing my lag which didnt happen before.

 

[johnbl]

you might also check your BIOS to see if there is a setting for USB. sometimes you can tell the bios to use a old version and turn off the USB extensions. motherboard vendors often attempt to speed up their usb by making extensions before the final specification is made. When the spec comes out the vendor has to update the BIOS and provide new drivers.
I think the 2.2 spec came out in 2012. you might see if there is a setting in bios to only use legacy or 1.x spec for the USB.
Problem is that newer drivers might not test properly with the old spec and you might just hit a bunch of bugs.

see if you can remove the device software or disable it. it stops responding after a while. you might have to use this method if the usbview can not remove the device (or windows PNP just reinstalls it)
https://technet.microsoft.com/en-us/library/cc771484(v=ws.10).aspx

Heres what the program says:

I think everything is fine though. I am just seeing a lot of ram usage going no where and i think thats what causing my lag which didnt happen before.

 

[whysarx]

fuck.

I just had another BSOD with cause being ATIKMDAG.sys. I will do a clean driver uninstall/install and see if it solves anything.

There goes the usb theory...