Unknown causes of BSOD in games, benchmarks and stress tests OK

[brekeke]

Please help detecting HW(?) issue.
CPU: intel core 2 duo 8400
former GPU: Asus ATI HD4850
new GPU: Sapphire HD7770
MB: MSI P35 Neo2

System is not overclocked, and never was.

A short history of the problem:
about a month ago, I experienced a BSOD playing Borderlands 2.
After restart I tried it again, got the crash within few minutes. After this, the card started showing artifacts and the windows would not even boot.
Next day, I was able to boot windows, but got the crash within GPU benchmark. After that, card was showing artifacts even in BIOS.

I suspected faulty graphics, so I bought a new one.

To my surprise, it started crashing the system just as before (BSOD), reporting either hardware error, or atikpmg.

I suspected drivers. Performed clean install of the system (Windows 7 64-bit).

I tried 13.12, and 13.9 and 13.4 drivers. No change in the behavior.

The problem is not limited to Borderlands 2, I tried X-Com Enemy Within (not really resource-hungry game). It crashes within 10 to 40 minutes.

Problem is that I can not reproduce the crash with any benchmarks.

I run memtest (bootable USB) for 4 hours. No errors.
I tried Prime95 and Intelburn for CPU test: no errors.

I tried Cinebench, Furmark, OCCTPT and "Video card stability test".
I also tried the OCCTPT and IntelBurn running together. No errors, no crashes.

I even ran HDD check. I ran OCCTPT power supply stress test.

I GOT NO ERRORS IN ANY OF THIS.
I really do not understand what is happening nor how to proceed.
Please help.

 

[John Connor]

Try running whocrashed and see what the crash is related to.

 

[denis280]

former GPU: Asus ATI HD4850 new GPU: Sapphire HD7770

Was it doing this with the Asus.

atikpmg.

as mention by John.get whocrashed.and also what brand and Watts is the psu.

 

[brekeke]

Whocrashed reported (for crashes both with Asus and now with Sapphire) either ntoskrnl.exe (WHEA_UNCORRECTABLE_ERROR), or problems with atikmpag.sys.

Now for the recent update:
The HDD I have is divided to two partitions: 80GB for system, rest for the data.

I have run Windows chkdsk first for the system partition - it went fine without any problems.

When I booted Windows, I have run the chkdsk from the GUI for the data partition (this did not require rescheduling after the restart - there are no system files on the data partition), but I did check "Fix errors on the volume", so it was taking a long time.

After couple of hours, I went to a crashed system - no BSOD, but corrupted graphics (like divided to 40 or so small tiles, accompanied with graphical artifacts and shimmerings.

After restart, the BIOS did not locate any SATA devices (this HDD and DVD/RW drive).

I disabled SATA controller in the BIOS, turned the PC off and after that unplugged from the power.
After that I plugged it back to power, turned it back on, enabled SATA controller in the BIOS and was able to boot to Windows.

I am now running the chkdsk from the command prompt again to see if it fails again.

A bad HDD? Or bad SATA controller (bad motherboard?)?

 

[Steltek]

The only way to tell for sure is to pull the hard drive from the system and re-run the hard drive diagnostic tests on a known good system. I would always suspect the hard drive first over a motherboard failure, but it pays (in terms of frustration) not to assume on any system over several years old.

Will the system still try to boot if you unplug the hard drive? If so, I'd download a bootable ISO from the hard drive manufacturer and run a set of extended tests on it. I've seen plenty of hard drives that go bad without triggering a SMART flag over the years.

Another thing to try is to swap the SATA cable - sometimes a questionable one will suddenly go bad.

 

[brekeke]

The only way to tell for sure is to pull the hard drive from the system and re-run the hard drive diagnostic tests on a known good system. I would always suspect the hard drive first over a motherboard failure, but it pays (in terms of frustration) not to assume on any system over several years old.

Will the system still try to boot if you unplug the hard drive? If so, I'd download a bootable ISO from the hard drive manufacturer and run a set of extended tests on it. I've seen plenty of hard drives that go bad without triggering a SMART flag over the years.

Another thing to try is to swap the SATA cable - sometimes a questionable one will suddenly go bad.

So, the HDD was probably defective - it hang up a windows chckdsk in a different system (had to reset the computer, not responding but no BSOD).

So I used Spinrite to test this HDD - at around 51% capacity, it completely disappeared (like the hdd was unplugged).

So I bought a new HDD to replace the old one, reinstalled the system on the fresh HDD (SSD).

BUT! I tried the XCOM game again, and while it took a bit loner to crash, it did indeed crash. (WHEA Uncorrectable Error 0x00000124)

Temperatures in stress testing are around 50C (max) for CPU, 60C for the new graphics card, 32C for the HDD, 35C for the motherboard.

I am totally out of ideas.

 

[denis280]

Please help detecting HW(?) issue. CPU: intel core 2 duo 8400 GPU: Sapphire HD7770 MB: MSI P35 Neo2

What psu do you have.and also what brand and model is you ram.and the ssd and hdd too

 

[brekeke]

What psu do you have.and also what brand and model is you ram.and the ssd and hdd too

PSU is Fortron FSP400-GLN 400W Blue Storm II

RAM: A-Data Vitesta EE kit 2x 2GB (running with default SPD timing)

Original HDD Samsung SpinPoint F3 3.5" 1TB HD103SJ

New SSD: Crucial M500 120 GB

 

[VirtualLarry]

If the HDD and SSD error, and then aren't being detected (no SATA detected at all), then that sounds like a motherboard or a PSU issue, honestly.

 

[brekeke]

If the HDD and SSD error, and then aren't being detected (no SATA detected at all), then that sounds like a motherboard or a PSU issue, honestly.

You misunderstood. The Samsung HDD was producing this kind of behavior ("disconnecting" from SATA controller), the new SSD is not.

The old HDD is now out of the PC, only the SSD is connected.

But the crashes are still happening, WhoCrashed reports either WHEA_UNCORRECTABLE_ERROR (caused by ntoskrnl.exe) or VIDEO_TDR_ERROR (caused by atikmpg.sys).

 

[jolancer]

dunno but a random thought.. you using any surge protector?

you verified an issue by replacing gfx, then u verifyed another issue swaping and then replacing HDD, sounds like theres more then one component thats being affected here, perhaps a surge occured, or possibly a lethal fault from a dying PSU. I dunno, but if it were me i wouldn't power it on untill i had a nown good PSU to test with and a surge protector if its not already.

 

[brekeke]

dunno but a random thought.. you using any surge protector?

you verified an issue by replacing gfx, then u verifyed another issue swaping and then replacing HDD, sounds like theres more then one component thats being affected here, perhaps a surge occured, or possibly a lethal fault from a dying PSU. I dunno, but if it were me i wouldn't power it on untill i had a nown good PSU to test with and a surge protector if its not already.

Yes, I use surge protection.

As for the PSU, weird thing is that if I run stress tests on both CPU and GPU simultaneously (i.e. arguably creating the majority of the power consumption), it can run just fine for hours and hours.

In the games, it can crash anytime between virtually seconds after the start to around 40 minutes.

 

[jolancer]

If the issue is isolated now only to your games, perhaps try running your game with audio disabled completely through device manager(or just pull the card if its not onboard). reboot and see if you then encounter any issues with the game

 

[brekeke]

If the issue is isolated now only to your games, perhaps try running your game with audio disabled completely through device manager(or just pull the card if its not onboard). reboot and see if you then encounter any issues with the game

Thanks for the tip. I disabled the integrated audio chip in BIOS, removed the Realtek drivers.
I even used Catalyst Uninstall utility, and reinstalled Catalyst drivers without the sound controller.
The game still crashes (though, interestingly, last two crashes only crashed to Windows, they did not produce BSOD)

 

[brekeke]

Well, newsflash: it just freezed in Windows, suprisingly with a similar behavior as with the old Samsung drive. After the resest, the SATA hard drive (the Crucial SSD) was not even listed in the BIOS (but the SATA DVD drive was).

I think it is the motherboard

 

[jolancer]

seems most of your components have encountered some sorta fault now. maybe worth while to replace PSU along with it, instead of risking PSU having been the root cause and damaging the new board.

or Fuk it all and wait till u can update your entire system, if u wana be more economical that way. but that road would require avoiding high load tasks such as gaming/StressTests/ExcessiveMultitasking, in order to keep your current system functioning. Assuming you were under load when windows itself crashed on you this last time