Build Crashing - ideas?

nozzferrahhtoo

My build is crashing a lot since wednesday. Nothing I know of has changed in the last few days.

Mostly it is the blue screen of death but every time the window 10 blue screen gets to 100% "information gathering" it does not reboot. So there is little information and no information in the Event Viewer either.

The code words it asks me to look up on the internet are different EVERY time however. I have not seen the one error code twice. One time I might get "irql_not_less_or_equal" and another time "bad_pool_data" and another time something else entirely about exceptions that were not handled.

Ran disk check on my drives, no errors. Ran memory test in my bios no errors.

USUALLY when I restart it just goes back into windows then crashes again 10 minutes later. ONCE however on rebotting it would not even get to the bios. Just kept restarting over and over until I unplugged it and plugged it back in again. That time I got some message from the bios that "over clocking had failed" but I do not over clock!

So I am at a loss, though it does feel like it must be something hardware related. Any ideas anyone? Or any good diagnostic softwares to recommend?

ED E

Sounds like the CPU is unstable.

Boot it up, try run Prime95 or Linpack. If that kills it its likely CPU(small chance ram). You need to check its not overheating(hitting 90*C) then its on to:
Power delivery from the board
or Degraded CPU

The latter I've never seen happen. The first thing I'd try to fix it there would be to disable any eco features in the bios, after that try upping the core voltage slightly.

Skatedude

Wait, win 10? the build listed 7 i thought, However, there was a big win 10 update a few days ago, Might be a suspect. Check the win 10 forums to see if others have similer issues?

K.O.Kiki

See if you can create a bootable Memtest86 http://www.memtest86.com/ USB drive and run a full test.

almorris

Hiya. I installed new ram last week and upgraded win 10. Same problems as OP is having. Just went into a loop of error msg's and wouldn't run at all. Ended up reinstalling win 10. Running fine since. Might not be hardware at all. See if you can role back any upgrades with win 10 first.

nozzferrahhtoo

Hmmmm I ran sfc /scannow and it fixed some dll called opencl.dll

There has not been a problem since but it has only been a few hours so I will give it awhile and see. If it happens again I will try some of the tools you guys recommended above, thanks!!

Skatedude wrote: »

Wait, win 10? the build listed 7 i thought

Yeah I upgraded windows when 10 first came out.

Cuddlesworth

K.O.Kiki wrote: »

See if you can create a bootable Memtest86 http://www.memtest86.com/ USB drive and run a full test.

I prefer using stressapptest off a Live ISO then memtest. I've had a 24 hour run with memtest that missed what the above picked up in 20 seconds. Built by google to test their servers

Run it as per the below
stressapptest -W -s 3600

nozzferrahhtoo

Cuddlesworth wrote: »

I prefer using stressapptest off a Live ISO then memtest. I've had a 24 hour run with memtest that missed what the above picked up in 20 seconds. Built by google to test their servers

Run it as per the below
stressapptest -W -s 3600

Ok I have to plead dumb on this one. That stressapptest app runs on Android or unix or something? So how do I run it? Some bootable USB drive with a small UNIX OS on it or something? Any recommendations?

The problem went away entirely for awhile after I ran SFC. But came back yesterday. I took the computer apart, cleaned it up (was super dusty), reapplied the Heat Paste to the CPU, and then put it all back together. It is now running 20 minutes without a crash.

But SFC is saying it has errors in some files that it can not fix. Things like:

Cannot repair member file [l:12]"normidna.nlp" of mscorlib, version 10.0.10586.212, arch x86, versionScope neutral, pkt {l:8 b:b77a5c561934e089} in the store, hash mismatch

Cannot repair member file [l:21]"System.Web.Entity.dll" of System.Web.Entity, version 10.0.10586.0, arch msil, nonSxS, pkt {l:8 b:b77a5c561934e089} in the store, hash mismatch

00005746 Hashes for file member \??\C:\Windows\diagnostics\system\UsbCore\TS_UsbCore.ps1 do not match actual file [l:14]"TS_UsbCore.ps1" :

00005702 [SR] Cannot repair member file [l:14]"dialer.exe.mui" of Microsoft-Windows-TAPICore.Resources, version 10.0.10586.0, arch amd64, culture [l:5]"en-GB", nonSxS, pkt {l:8 b:31bf3856ad364e35} in the store, hash mismatch

And so on and so on. About 30 or 40 files listed in the log.

K.O.Kiki

How about reinstalling Windows?

Drastic but might help.

Backup important files first.

Cuddlesworth

Cuddlesworth wrote: »

Create a linux mint live install on a usb using RUFUS

Install Google Stress App test by running this command in a terminal window "sudo apt-get install stresstestapp", then to run the app "stressapptest -W -s 3600".

Do the above, it will run for 30 minutes. You can learn a little about linux, its helpful for troubleshooting. If there is a memory issue, it will return a fail message in the terminal. Memtest has been underdeveloped for a while and isn't the best anymore but its still good to use in conjunction with the below.

http://google-opensource.blogspot.ie/2009/10/fighting-bad-memories-stressful.html

"This test was designed to test all memory available on a machine, which is not guaranteed with the execution of a CPU-intensive application (for instance, compiling the kernel on multiple threads). Moreover, it is focused on testing the memory interface and connections, not the memory internally, like memtest86"