Ghost in the Machine

[Questor]

I am seeking community help with fixing one of my rigs. I will do my best to keep this orderly.

After the machine somehow got a rootkit on it, of course the machine started flaking out. I used two tools from respected companies (Trend Micro and AVG) to ensure removal. Then, as the system was in such a horrid state that restore from back up wouldn't even work, I zero'ed the drives to DoD specs for extra safety and reinstalled Windows XP. Of course we now (the post is a dead give-away) know there is still something wrong.

The list:
Gigabyte M57SLI-S4 Rev 2.0 MB
AMD Brisbane Athlon 64 X2 5000+ CPU (2.6GHz) -
4 GB Corsair XMS DDR2 800 (PC6400) RAM
2 Seagate Barracuda 250 GB HD - run as single drives on SATA channel
2 Seagate Barracuda 80 GB HD - RAID 0
Corsair HX620 Modular PS
Using onboard sound and NIC
ASUS 9800 GTX video card - single
Lite-on DVDRW burner 20X - SATA
Windows XP Pro Media Edition OEM (It was free! - legally)
The rig is plugged into a backup power supply, so no hard shut downs.
Nothing is or was OCed

What happens:
The machine hangs leading to reset (the only thing that works to get it going again), crashes and reboots to a black screen and of will sometimes BSOD "out of the blue," or after a crash. Windows often calls itself to run CHKDSK after which, everything runs fine, for awhile (awhile equally a completely random amount of time)

Oddities - other than above:
One stick of the four Corsair XMS will not allow me to reinstall or run Windows, yet checks perfect with the latest Memtest. Contacted Corsair, got RMA, solved.
Any one of the other three sticks in the same MB RAM slot will allow flawless Win XP reinstall and the machine run perfectly, as far as the RAM goes.
Any drive I use frequently gets errors prompting Windows to again call for CHKDSK. Again, after which everything runs fine again for awhile.

What I have done:
Ran 2x Memtest on each stick, each matched (dual DDR) set installed separately and all four sticks installed. Result were flawless, yet the above mentioned oddity.
Ran SeaTools diagnostics. No error codes or any issues reported.
Tested the power supply, including each connector. You guessed it. The tests showed no problems there.
I used all new drivers for the reinstall of Windows. When that didn't seem to work - read: next unrestorable crash - I used earlier versions of ALL drivers, including video. It made no difference. The same random crash, hang, BOSD was observed. I updated the BIOS. There was no difference. So I reverted back to the earlier BIOS version (if it aint broke, don't fix it).
I have moved the for sure known good DDR set to slots 3 & 4. This has resulted in no change.
Let's see, changed out all the SATA cables, one by one. The result was no change.
It does not matter which of the four drives I use or any combination of them, in raid or as single drives. I get the same result.
Changed the power supply cable sets, not using some at all, then using just the previously non-used and then using all of them spread around.

Understand, when it runs correctly, this rig runs really nice. It's this darned ghost!

Yeah, okay, Motherboard or CPU?

Feedback is desperately requested

Best Regards to the Community,
Questors

 

[C1]

Your trying too hard. You have too many variables in the mix to isolate the problem. If you really want to pin it down then you have to tear everything back.
- Clean all the connectors to the MB & use Deoxit and reconnect (includes Video card PCI spade)
- Remove/Disconnect everything except the video card.
- Turn everything off in the BIOS (eg, LAN, Sound, USB, etc.)
- Boot from a DOS (eg, WINME startup disk) and run the system. Run Gorilla.BAS; Does the machine fault; if so cycle the memory sticks (one by one)
- If the machine still faults you may want to re-seat the CPU as a last resort


In any event, the strategy is to try to get the system peeled back to a state where it works without faulting then add parts back piece by piece till the problem is found.

For example, I encountered an intermittent CPU fault after vacuum the dust out of my machine. Turns out it was something in the main PSU connection to the MB that got disturbed & you would never have guessed it by looking at it (or even the error).

In fact, just had an issue TODAY WITH AN HDD CONNECTION. Damn-dest thing you could imagine. Trying to test the HDD & it would show/register on some boot ups & not others (drive/connectors undisturbed). Clean connections with R12 then Deoxit (both male & female) & all now works well/consistent.

 

[Questor]

Yes sir, the basics of PC troubleshooting. Go basic, eliminate and add back one by one. I guess after awhile we forget these things. Thank you for the suggestions.

-Questors