Strange Hardware Issue, trouble diagnosing.

[Kilyin]

Not sure when this started happening, and I've never seen anything like it before.

On a cold boot, my PC is perfectly stable. Passes memtest86+, passes Windows mem diagnostic that comes with Vista, and passed Windows memtest by hci. Also passes Orthos small fft.

After a day or two, it will fail Windows memtest by hci, and if I warm reboot at that point it will also fail memtest86+. It also will blue screen occasionally when it's in the unstable state... PFN_LIST_CORRUPT 4e stop error or KERNEL_MODE_EXCEPTION_NOT_HANDLED 8e stop error.

Now, the interesting part... if I power the PC completely off and do a cold boot, it is back to normal... passes all memtest86+, windows mem diagnostic, Windows memtest by hci, and Orthos.. but after a day or two, it's back to the unstable state. I noticed this because Azureus would crash all the time and IE would just close by itself. Then I got a BSOD. I almost RMA'd my RAM because it was failing the memtests in this state, and I had it down to 1 GB stick thinking the other was bad - because I had to power down to remove one stick. Of course that temporarily fixes the problem, so I thought I had figured it out and requested an RMA. Then the next day the single stick displayed the same symptoms. So I take it out (another cold boot) and put the other stick in by itself that I thought was bad, and then it passes all the memtests. So I just ended up putting both sticks back in.

I have never seen anything like this, and it's maddening because I have no idea what it could be and it takes 24-48 hours to reproduce. I have tried lowering my CPU overclock (Even though it's stable) and I have now disabled DRAM static read control and loosened the timings - will see tomorrow if it does it again. I also flashed to the latest bios for my P5K Premium (0612). I'm hoping it might be fixed, but my gut tells me otherwise.

Any ideas?!

Specs:
P5K Premium
E6850 @ 3.9 GHZ
G.SKILL 2GB (2 x 1GB) F2-6400PHU2-2GBHZ
8800 GTS G92
FSP 760w PSU

[Serpentarius]

maybe some loose screw or something ...

this is just a suggestion .. u may not use this ..

i suspect that your RAM might be loosy ... or something .. you could get an eraser (not pen eraser) .. and briefly, lightly scrub on the copper then place it back
sometimes ... misplaced RAM might cause probs .. i got a dozen of clients have the same probs ..

remember to place your casing in horizontal before slotting back the RAM ... it helps in getting the right accuracy ... and reboot bout dozen times or so ...

[SoulsCollective]

Sounds like something is building up a charge while the machine is on, which then discharges when you power the machine off.

Is this a new build, or has this been up and running for a while before developing these issues? Also, are you noticing any gradual decline in stability over the 24-48hr period?

[Kilyin]

I got the board, RAM, and PSU last September or so. I have done a few upgrades since then (like the processor), and the PC is watercooled. I wish I could pinpoint when it happened, but I haven't been using the PC intensively for awhile. I basically remote into it from work and browse the net and do some downloading at home. I haven't even played a game recently.

What could possibly be building up a charge?

Is it possible for a CPU to go bad, but still operate and appear to be functional? At this point, I have little choice but to start swapping out parts. Gonna have to buy some more DDR2.

[zanzabar]

can u try a diffrent psu, and waht ram cooling do u have

[Kilyin]

Don't think it's the PSU, all the rails read solid in the BIOS, and Fortron isn't known for dud PSUs. Unfortunately I don't think any of my spares are beefy enough to run the system, might buy a new one for testing.

I have a 92mm fan blowing on the RAM, but it doesn't even get warm without it.. I only run 2.2 volts to the RAM.

[XtremeBawls]

I used to have a 2.8 Pentium D that you would overclock, it would be completely stable, then a few days later, it wouldnt, so you'd drop the speeds and it would be completely stable, then it would do the same thing and just over and over and it would always be a few days before it would just decide to not run right. Found out it was faulty, got a new one, it was a gem too, it oc'ed like crazy, now I think it is in one of my friend's customer computers that he built for someone....such a shame, but I didn't want to pay for two rigs at the time i upgraded.:( (if i was to do it again, i would have kept it that's for sure)

[Kilyin]

Seems to be fixed now. Using all the same hardware. Changes I made were booting into a PE environment and deleting the paging file (can be done without PE but I was checking some other stuff), disabling DRAM Static Read Control, adjusting timings from 4-4-4-10 to 5-5-5-15, and flashing P5K Premium BIOS from 0204 to 0612. No more 'incidents' since the 16th and 24/7 operation. I've actually increased my OC on the CPU and RAM as well (clock speeds), so we can rule that out. I don't really care what fixed it, just happy it's not happening anymore.

Most bizarre computer problem I've ever encountered without running into a bad piece of hardware.