System passes all tests then randomly fails?

[Codesmith]

I am having a problem with a Asus K8V SE Deluxe system I build 2 and a half years ago.

The owner had some problems with spyware and virus infections.

I fully tested his hardware, reinstalled windows, but after three days the system crashed twice for no reason then failed to POST

The system has talking BIOS which reported CPU Test Failure over and over again.

I cleared the CMOS, swapped out/disconnected everything but the CPU/Motherboard, removed it from the case. Tested the other parts in another system.

Then it just started working again. I did blow away a tiny bit of dust near the CPU right before it started working, but I didn't buy the dust theory.

Nothing at all in the even log about the crash. Its as if someone just yanked the CPU out and then put it back in.

I ran memtest 86 for 12 hours, prime 95 another 12, then ran the following all at the same time for another 12 hours.

Prime 95
Sandra Burn In Wizzard (50% max CPU)
Video Card Stability Test 1.0.0.2
Western Digital Diagnostic on the hard drive.
Nero CD/DVD speed on the optical drive.
I even transfered some files to get the ethernet card working.

Also ran Sandra's environmental monitor the entire time and all the voltages and temps were within normal range. I even tested the PSU with a multimeter because motherboard sensors are not that accurate.

I finished installing software. Then a week after the first incident the owner calls me and says it was running overnight and now it says its saying CPU Test Failure again.

This time it corrected itself after a reboot. And now its working for a few days since.

----
So it passes all the tests I throw at it, works under full load for days at a time then randomly cuts out and stops recognizing the CPU for brief periods.

Sometimes it won't post for half an hour despite all attempts, sometimes it posts after pressing the reset button.

The only two parts that were always in the system when it failed to post are the CPU and motherboard. I don't have another socket 754 system to swap CPU's with.

I can't do much trouble shooting because the system works 99.9% of the time and doesn't have any symptoms to suggest a problems while it working.

I am at a loss what to do next other than try RMA-ing the motherboard to see if that fixes the problem.

The owner is happy with the system as it is as long as it doesn't get worse. Plus he would rather wait to see if whatever part is bad will fail completely so we are 100% certain it needs replaced.

---
Anyway I am wondering what people think and whether there are any tests I left out, or things I haven't tired.

 

[_Morphine_]

First thing that comes to mind is a broken pin on the cpu.

 

[Codesmith]

Yeah I checked the CMOS battery when I tested the PSU with my multimeter.

Can CPU pins just break? Because I definitely didn't bend/break any pins when I build the system 2.5 years ago.

===

I found out that the owner has a metal shop in his basement, and sometimes does some metal grinding.

So maybe I did fix the hardware stability problem when I dusted the motherboard. There wasn't much dust but if some of it was metallic it wouldn't take much to cause problems.

It hasn't crashed again so that about two weeks running stable.

The one crash that happened after I dusted it could have been a fluke.

 

[_Morphine_]

I have heard fluke problems fixed with dusting before. At that time the theory was that the dust had become magnitized and created temporary ESD interference.

FYI, a CPU pin wont just break unless there was some instance you remember a rough removal or storage environment etc.

It reminds me of the time when I was a kid driving a forklift and I heard a strange crunch and figured I ran over a water bottle. 3 hours later while looking for my sunglasses I put 2 and 2 together.

I could imagine a broken pin still in place that would make contact then lose it creating a difficult troubleshoot, since you seemed to have tried it all.

Thats an interesting problem though with the dust.