Hey guys. Long time reader first time poster. I apologize in advance for any missed practices.
I read up a lot on this forum before building my custom rig back in December. I've been running everything on default. Didn't feel a need to Overclock so never did. Most of my schoolwork has something to do with VMs so I opted for the 2600 not the 2600k for the Intel Virtaulization Technology. I use my 60 SSD as a boot drive with my users and other folders moved off to my 1TB SATA. The SSD simply hosts my necessary Windows files and some major programs.
Everything worked fine up until this past Sunday. Had a Google+ Hangout going as well as my VirtualBox CentOS. I heard a static buzz through my headset as my computer froze up and then restarted.
After that, I started having BSODs and more frequent freeze/restarts. The worst situation consisted of a BSOD every 5-7 mins or so. Most of the BSODs were either "IRQL NOT LESS OR EQUAL" or the following image.
After the CPU0 interrupt, I decided to run memtest86+.
I did it 1by1 with all sticks in slot A1. I then proceeded to do 2x2 with A1/B1. I did then 2x2 with A2/B2. I finished it up a full 4x4. All of these tests passed without any problems.
Today, I went ahead and started a prime95 torture test on my machine. It got about 5 mins in with all CPUs at full capacity and only running at 45-47Cs when it would restart my computer. This happened on 3 different occasions each within the first 5-7 mins.
Obviously, something is going wrong with my machine. I didn't install any new software or applications. I did a BIOS flash after the original BSODs. I checked to see if drivers were up to date and they were. I ran CC Cleaner and fixed any registry issues that showed up.
The Prime95s were ran before and after the above attempted fixes. The computers been running idle now for about 3 hours just fine, but that was the case on Sunday night as well.
I would appreciate any help and guidance that you guys could provide.
Not sure if this is the right section of the forums for this. If not, could someone please move it or direct me to where I should post it.
Computer Specs Follow:
Intel i7 2600 @ 3.4 GHz
ASUS p8z68-v LE mobo
Patriot Sector 5 G Series 4x4 GB DDR3 XMP Ready (Running at 1333 MHz, 9-9-9-24-2, 1.5 V)
OCZ Agility 3 60 GB SSD
Seagate Barracuda (ST310005N1A1AS-RK) 1TB 7200 RPM
ASUS HD 6570 1GB DDR3
Corsair H80 CPU Cooler
Antec Earthwatts 650 Watt PSU
NZXT M59 Case
Windows 7 Professional SP1
Though you managed to get "passes" in memtest, the fact you got loads of errors makes me think RAM (or at least in the RAM bus). When you did get passes, did you let memtest run a single pass, or let it run for hours (i.e., 6-8hrs)? Unfortunately it can take memtest a good few hours (several passes) to expose a small problem.
Unfortunately if it's not RAM, I suspect the memory controller or traces on the motherboard. RAM is the easy thing to RMA (should you not want to go through the hours of testing them individually in memtest).
The next step I'd take is the CPU because the memory controller is inside that guy (and recommending in 2nd place only because it tends to be easier to RMA a CPU than a mobo - mobo takes a lot of work to harvest out of a case)
Lastly is the mobo, but admittedly, I suspect this has a higher probability of being bad vs the CPU.
If you have available to you another hard drive (so you don't have to wipe your existing config), you could try reinstalling Windows to insure you don't have a driver corruption. But I doubt that's the case given the memtest results you have.
None the less, should you go down this path, reset CMOS properly, and set your BIOS settings to the non-OC'd settings you were running this system at. Then do your reinstall (also insuring AHCI is enabled and you're using the right Intel drivers). Again, I do suspect this test cycle is going to produce BSODs still, but if RMAing parts is troublesome in your region, then this might be a worthwhile test.
RAM / BIOS:
F5 to load Defaults
Memory Frequency -> DDR3-1333
DRAM Timing Control /Enter
DRAM CAS Latency -> 9
DRAM RAS to CAS -> 9
DRAM RAS PRE -> 9
DRAM RAS ACT -> 24
DRAM Command Mode -> 2
VCCIO Voltage -> 1.15v~1.20v
DRAM Voltage -> 1.55v ; (2) kits
F10 Save = Yes
Then test Memtest a minimum of 4-passes and best overnight.
As far as testing the MOBO and CPU, replace one of the other. There's no 'real' testing applications. There is ASUS PC Diagnostics but it's worthless and I'd dry AIDA64 Extreme Edition which digs pretty deep; see - http://www.aida64.com/downloads
Try the modified memory settings - first. If all is well then Run 'Fix It' and Schedule SFC both options...
When I ran Small FFTs, the test ran without any errors or warnings for a good hour. My CPU temps hovered around 48-50 Celsius. This test only used about 2GBs of ram which was ~14-16% of my total.
I then stopped and ran a Blend Test. The test put my RAM usage to just over 4GBs and ~25%. Within a minute of starting the test, my PC had restarted itself. The temps were the same as in the Small FFTs test.
If after modifying the BIOS as I posted above you have any errors in Memtest then you'll need to test (1) stick at a time. CPU: | blank | 4GB | blank | blank | ; pass then swap the stick. I'm assuming you have (1) bad stick...
Run Memtest first and with ALL sticks installed, based upon the photos they should show-up quickly (Test #1), and with the BIOS changes posted above.
All the errors that showed up were in my original three runs with Memtest86. I then proceeded to run all sorts of variations including an 8 hour overnight run of Memtest86+. None of my 86+ tests reported any errors.
With prime95 still crashing, there's obviously something wrong. Should I just RMA the RAM and if the problem still exists proceed to RMA the CPU/Mobo?
One last go, some of the BIOS versions in March were lets say less than 'stellar' so I am assuming you have BIOS v 3702, try downgrading 1 notch to v 0805.
Often when eliminating to OS i.e. running Memtest is okay then it 'can' be an OS corruption. Run MSCONFIG and select Diagnostic, reboot, and try Prime95 again.
Failure, remover the GPU and unplug any headers going to the Case and try once again.
If it comes down to it possibilities: PSU, MOBO and CPU. PSU only seems unlikely because of the Memtest failures; whereas Prime95 spikes the load to the PSU. Therefore, the statistical order otherwise is: 80% MOBO then 20% CPU -- if you perform a full breadboard first with bare minimum configuration.