BSOD when gaming (even after many fix attempts)

[Eldiran]

Hey, I've got a longstanding BSOD issue that I just can't get rid of. I'm getting quite desperate.

Simply put, I get BSODs when playing any game that is remotely hard on the computer. Left 4 Dead 2, Skyrim, Chivalry, Terraria, Starbound (when hosting a Terraria/Starbound server on the same computer), and Titanfall all give me BSODs after 5 minutes - 2 hours of gameplay. (Generally the more intense the game, the sooner it crashes, but it varies wildly.) The BSODs occur even if all graphics settings are set to Low.

I never BSOD when not gaming.

The error is an "uncorrectable hardware error". WhoCrashed points to hal.dll and gives no other useful info. None of my temperatures exceed normal values (the highest is my core, which reaches 65 degrees). Memtest comes back with no errors. SeaTools tests all pass.

The problem has been going on for over a year now. Since it began I have reformatted the computer, replaced the power supply, replaced the motherboard, and replaced the graphics card. None of that affected the problem.

All drivers are up to date (I bought all these brand new parts very recently, to try to fix the problem).

I can accept that I need to replace more parts. However, I don't know what could possibly be causing it. I'd settle for a good lead on what to replace next.

Please help!

EDIT: forgot to mention, the computer is not overclocked.

 

[Thedoctor44]

Right, because you have bugchecks that consist of 0x124 errors (WHEA_UNCORRECTABLE_ERROR) which stands for Windows Hardware Error Architecture. This means, as you can probably tell that it is a hardware error within the computer. I have encountered issues where drivers are the cause which in fact has happened to me but it is very unlikely.

we should begin with a few stress tests. How many passes has memtest passed for? You must run it for at least 8 passes.

Stress test your CPU by running prime95

www.sevenforums.com/tutorials/100352-hardware-stress-test-prime95.html

Stress test your GPU with Furmark

http://www.sevenforums.com/tutorials/100356-video-card-stress-test-furmark.html

Post the results.

 

[Eldiran]

Thanks for answering; I was afraid I wouldn't get any responses.

Memtest ran for more than 8 passes.

I've run Prime95 before with no results, but I ran it again for an hour to be sure. Didn't get any errors or crashing.

Ran Furmark for 9 minutes with no results. Temperature topped out at 90 degrees with no crashing (also, when gaming, I rarely exceed 65 degrees according to Speedfan). It shouldn't be possible that the GPU is a problem anyway, because I recently replaced my graphics card, and it didn't affect the BSODs.

 

[Thedoctor44]

Hmm, on rare occasions, drivers can cause 0x124 errors. I've encountered them personally (unfortunately). Can you enable Driver Verifier to see if a faulty driver can be identified.
Follow these instructions.
http://www.sevenforums.com/tutorials/101379-driver-verifier-enable-disable.html

 

[Eldiran]

Alright, followed the procedure, then got a crash. Attached are the dump files.

I shouldn't think it would be driver issues, unless I'm super unlucky, considering I've replaced most of the parts with drivers (Mobo/GPU)... but you never know.

I also caved and bought a new hard drive, so when that arrives in a few days I can find out if the HD was the cause. (At least it was an upgrade...)

 

[Thedoctor44]

The crash analysis is inconclusive, however it points to Atikmdag.sys which is the AMD GPU. The other bugchecks pointed to hardware so I'm pretty confident the cause is graphics card related.

 

[Eldiran]

Really? I already replaced my GPU... : ( admittedly both of them are AMD.

As a test I ran AMD's driver cleanup, then installed the absolute latest (beta) AMD driver set. Still crashes.

I didn't always have this crash issue with my old graphics card. Do I need to get legacy drivers or something...? Or maybe I could get just the driver and not all the extra stuff like AMD Catalyst Control Center?

 

[Thedoctor44]

I'm afraid I can't vouch for AMD, I've never had any of their products. You could try installing a graphics driver without any extras.
But it does indicate a GPU hardware error.

 

[Eldiran]

Yeah, removing all the extras, like CCC, did nothing.

This is ridiculous. It has to be something pre-existing (i.e. not a result of flaws in my brand new parts), or it wouldn't have persisted across all combinations of 2 mobos and 2 GPUs. The only unchanged parts are the CPU, RAM, and the HD.

Aaaaand now my BIOS is freezing up. So I can't even tweak the settings to see if they are the cause.

Thanks for the help, by the way. I might still not know anything about what is causing it, but this is narrowing down the possibilites, which will likely be very useful.

 

[Thedoctor44]

Hmm right lets think. It's not software because Windows hasn't loaded up yet, so drivers and other processes can't be an issue.
Correct me if I'm wrong:

• Prime95 ran for at least one hour with no errors or crashes whilst running the application.
• Furmark ran for at least 30 minutes with no artifacts on screen or crashes whilst running.
• Memtest86 ran for at least 8 passes with no errors.
• SeaTools ran a test. (Note, you can try running a long generic test just to make sure).

The main thing I can think of now is a faulty motherboard. However it is a long shot and I'll ask for someone with more knowledge for this situation.

 

[Eldiran]

Yep, you summed it up right. I've also monitored all temps and found nothing overheating even when the crashes occur. Motherboard is a long shot for the BSOD issue, because it'd have to be 2 separate mobos (different model) with the same problem in a row.

The bios problem could be unrelated, so I don't know what to make of it...

 

[Thedoctor44]

You've also enabled Driver Verifier for at least 24 hours and it found nothing right?
If not then please enable it by following these instructions.

http://www.sevenforums.com/tutorials/101379-driver-verifier-enable-disable.html

 

[Eldiran]

Oh, I only enabled it for 1 crash. I thought it improved the crash log somehow. What is the benefit of keeping it on for 24 hours? Does it create a special diagnosis after a certain point?

 

[Golden]

The purpose behind running Driver Verifier for 24hrs is to give it sufficient time to completely stress all the drivers, maximizing its ability to cause a BSOD in which the faulting driver can be identified.

If you receive a BSOD shortly after using the PC with Driver Verifier enabled, then it has done its job. You can then disable it, and upload the .dmp files using this tool so it can be analysed:

DM Log Collector.exe

 

[Eldiran]

Ahhh, thanks for the explanation. I certainly don't need the help in generating BSODs : P

I don't know if there's any additional info here that isn't attached to the first post, but just in case, I've attached the latest dm log, which should also include the crash caused with Driver Verifier.

 

[Eldiran]

Update: the new hard drive didn't fix it. : (

 

[Thedoctor44]

Well these 0x124 bugchecks indicate the CPU is at fault but you've already ran Prime95. Please run it again for around 2 hours.

===============================================================================
Common Platform Error Record @ fffffa8005223028
-------------------------------------------------------------------------------
Record Id     : 01cf4dfda1fb8de4
Severity      : Fatal (1)
Length        : 928
Creator       : Microsoft
Notify Type   : [COLOR="red"]Machine Check Exception[/COLOR]
Timestamp     : 4/1/2014 23:12:58
Flags         : 0x00000000

===============================================================================
Section 0     : Processor Generic
-------------------------------------------------------------------------------
Descriptor    @ fffffa80052230a8
Section       @ fffffa8005223180
Offset        : 344
Length        : 192
Flags         : 0x00000001 Primary
Severity      : Fatal

Proc. Type    : x86/x64
Instr. Set    : x64
Error Type    : [COLOR="Red"]Cache error[/COLOR]
Operation     : Data Read
Flags         : 0x00
Level         : 1
CPU Version   : 0x0000000000100f43
Processor ID  : 0x0000000000000002
Instr Pointer : 0x0x000000006741f162

 

[Eldiran]

Uuugh... what the heck...! Prime95 tested perfectly before, but now I get crashes within a few minutes! Maybe the behavior changed because of all the replaced parts. I don't even know.

Anyway, the 'good' news is Prime95 crashes real quick. When I run 'Blend' tests, I get a BSOD "uncorrectable hardware error" -- the same I get when gaming -- within 2-5 minutes. This is easily replicatable with Blend tests. The bugcheck code is 0x00000124. Whocrashed points toward hal.dll for these.

When I run 'Short' tests, in 20+ minutes (sometimes it lasts a while, sometimes it happens quite quickly) I get a BSOD saying "clock interrupt was not received on a secondary processor within the allocated time interval." Error 0x00000101. Whocrashed says the error is either hal.dll or ntoskrnl.exe for this kind of crash.

So... hopefully this is useful information for diagnosis. I just wish it had shown up earlier. At least all the part replacements I tried were upgrades...

 

[Thedoctor44]

Right, that definitely sounds like a CPU issue, I would highly recommend replacing it.
I'm sorry that you had to replace lots of parts but unfortunately that ends up being the best route to go down when dealing with 0x124 bugchecks, it's mostly trial and error.

 

[Eldiran]

Yeahhh, I've gone ahead and ordered a new one. It's actually a good thing I replaced my motherboard for one that supports AM3+, however, since apparently AM3 CPUs are being discontinued (couldn't find any non-used ones for sale, anyway).

The old parts aren't going to waste regardless, so it's cool.

I'll post an update when I get the new CPU.