New computer - constant BSOD - baffled as to root cause

[Monomakh]

Hi all,

I just built my new system (I've built many over the years, never a problem). The new PC was great for about 1 week (including a 24hour burn in), then all of a sudden starting yesterday night, I began receiving BSODs constantly. This has been extremely frustrating and I'm desperate for answers.

Hardware (all brand new):
Windows 7 Ult x64 OEM
i5-3570k
16GB G.Skill PC1600
EVGA Nvidia GeForce 670
Corsair 256G SSD, 1.5TB SATA hdds
Corsair 750W Pro silver PSU
Antec 302, CoolerMaster 212 heatsink, 3x case fans
All drivers were the latest updates. BIOS updated to latest before installing OS the first time.

As my system is currently in pieces across my workstation and I'm typing this on my Mac, I don't have access to many of the dump files from the crashes, though I do have 5 available if they would be of any use. The errors were all of the "caused by Address ntoskrnl.exe+7f1c0" type, with BSOD messages of MEMORY_MANAGMENT, IRQL_etc, or various others that typically seem to point to drivers or RAM.

Behavior: system ran great for a week. I tested various OC settings, nothing extreme (e.g. CPU @ 3.5GHz instead of 3.4) and went with standard settings for the main (the OC actually decreased video card performance). Midday Monday (yesterday), the Windows update autoinstalled some packages, possibly in the middle of auto system backup. I returned home to find the "your system has quit unexpectedly, check for solutions dialog." Checked, no solutions. I launched a browser and... boom. BSOD. Reboot. BSOD on Windows startup. etc. I ended up removing every bit of extraneous software and drivers I could find and it seemed to be more stable - I played Battlefield3 (the real reason I bought the machine, let's be honest) ... for about 40 minutes, then it crashed. After the BF3 crash, the system was not able to stay functional for more than 5 minutes at a time. Dump logs analysis was inconclusive. Heat tests post-BF3 were within spec - CPU/GPU ~41C or less. Ran some hardware tests, typically from boot-USB or safe mode as system was too unstable to stay in Windows.

Giving up on attempting to fix via software, I thought a clean install might fix the issue (which seemed to be driver-related at the time). I blew up the Windows install drive, deleted the partitions, and went to reinstall Windows. Only to find that now the Windows installer will not successfully complete installation and hard freezes after the 'expanding files' section. *However* the system was entirely stable when I started the installation, and left for work, coming back 10 hours later with the machine still on and waiting on the disk selection screen (so at least we know it was stable at true idle, no activity other than displaying VGA screen). Also of note: during these installs, the Windows install would not recognize my USB mouse (though it did USB keyboard) which was annoying and also different than the first install (where mouse was recognized). After the complete teardown, rebuild, and attempt to install Windows this evening, the mouse was recognized by the Windows installer. This, to me, pointed at the mobo as well.

I even put a Linux live CD (Mint 12, in case it matters) to test stability and the system hard locks after maybe 2 minutes post-getting-to-desktop.

As this is by and large the same behavior that just killed my previous machine (3yo), where I first thought it was the RAM, but turned out to be the mobo frying the RAM (... and the replacement sticks I put in), I checked the RAM first for hardware issues, but they passed memtest on multiple iterations over multiple hours.

Hardware diagnostics/tests:
* Ran memtest more than 10 times. No errors found in either stick.
* Checkdisk reported no errors.
* SFV returned corrupt system files once, but irrelevant with later clean install attempts.
* No overclocking.
* Temperatures all within norms (~32C at idle).
* Voltages stable
* took machine completely apart. Checked every connection, including screws. Disconnected every nonessential component, leaving only fans, SSD, and DVD drive. All POST LEDs on the motherboard reported no errors.

Since every piece of hardware passed, I RMAd the motherboard (shipping it back tomorrow) and hope the replacement fixes these crashes... and I get it before the weekend so I can actually spend time on the system. If that doesn't work, my fallbacks are RAM, then the SSD, but both seem remote. My last mobo was Asus as well, but should I try another brand? I thought the premium pricing would help avoid these problems (if it is the mobo), but identical behavior on 2 systems in a row built 3 years apart tells me I may have backed the wrong horse.

Have I missed anything? What should I be looking for? I'm at a loss as to where to look next if the new mobo does not fix things, though I've of half a mind to return everything and live in my no-gaming-allowed OS X Siberia for another year.

All help is most appreciated.

 

[deepb]

Did you try installing the OS on your 1.5TB Hard drive instead of the SSD ? If the windows install is not being completed 3 possible problems - SSD, Motherboard or Installation Media/DVD Drive. Since you would have already eliminated your motherboard issue by getting a replacement, you are left with eliminating the other 2 possibilities.

Sounds like you are just having bad luck. That's quite a system build you have got there. Keep at it and you will have tons of fun gaming on that.

 

[Monomakh]

Thanks. Since Windows installed just fine the first time on the SSD (and lasted for a week)... I did not try installing on the hdd. It's a good idea, I simply did not think of trying that.

To clarify a bit on the install fail - the files are copied to the SSD and expanded just fine, it's when Windows reboots out of that install screen to the first boot of the Windows graphical environment to finish the install that the system goes to the hard BSOD. So far as I can tell, the drive itself is not failing (or the media or DVD drive, given the same behavior on Linux disc and previous successful installs). I guess it could still be bad RAM too, if the memtest gave me a false negative error result.

It's the Linux lock up that was the it's-definitely-the-hardware kicker for me (clearly, no shared drivers with the Win side), I just cannot imagine a system fail out of the blue like that, post-burn in (especially with my way-overbuilt machine) after minimal usage. No thunderstorms or obvious power surges. I hope it is just bad luck, never had this happen in 'lo my many years of system building. Damn gremlins.

 

[deepb]

If it's failing on first boot, could be the 670 or Ram as well. You haven't mentioned your motherboard model, other than it being an ASUS. If it's a got integrated graphics support, try running without the 670 connected to eliminate something being wrong with the GPU. Also, the Ram ( I'm guessing it's 4x4 gb configuration) you could try just running with 1 of them and see if that helps.

Let's hope you won't have to troubleshoot at all once your motherboard arrives. Good Luck.

 

[Monomakh]

ASUS Z77 Sabertooth = the mobo model, which is supposed to be "military grade" tough. Can't believe I forgot to mention that above. My bad.

On the RAM - that would be the next likely culprit (it's 2x8), despite passing the memtest+. Question: should I order new RAM preemptively (or RMA) ... or wait for the mobo and see if it still fails?

Is there any program that can give good diagnostics on the GPU? I really hope it's not that (or the CPU itself), but I also don't know how to eliminate it with certainty.

 

[Monomakh]

I just mailed the mobo back for RMA today. I assume 2 days shipping. Even with same day review and return on the RMA, I'm looking at Monday at the earliest for the motherboard, which is annoying as I lose the weekend (and my only real free time) and I'll be distracted until I solve this problem.

I've been debating ordering a different mobo (MSI Z77A-G45 probably) so I can get it before the weekend and see if I can change the RMA to a return instead... except I just realized I did not include all of the SATA cables or the case connector cutout thing in the RMA box. Dangit.

I haven't RMA'd the RAM due to it passing the memtest, but I still don't trust it. I could order new RAM and have it in by tomorrow or Friday. Or, play RAM roulette with the 6 sticks (4x4GB, 2x8GB) I have sitting around, of which I know 1 is definitely bad (but can't remember which one) and am suspicious about all the rest given current events. Then again, for $70 for high speed 8GB I'm not sure it's worth rolling the dice and waiting either.

 

[Monomakh]

Nope, I have no motherboard so I'm at a standstill - it is in the mail to Newegg for the RMA.

 

[Monomakh]

Alright, rush ordered a new motherboard (Asrock Extreme4) and ... it wasn't the motherboard. Hard BSOD on Windows install.

I then took out one of the sticks of RAM and... it was the (tested as error-free) RAM. Took out one of the sticks and the system has now been running for 12 hours with no crashes. Knock wood, but I think that did it.

Of course, now I've got an extra motherboard for no reason whatsoever. Pretty much just set my money on fire there.

 

[deepb]

E-bay that spare motherboard and recover some of the money. I'm sure a Z77 with warranty intact shouldn't net you too much of a loss. Have fun gaming in the meanwhile .