Catastrophic system meltdown: "Testing Memory 4193"??

[manveruppd]

After a system freeze which resulted in a black screen (not a BSOD, a completely blank screen with no display), I hard-reset the computer. During POST, I noticed that it had reset my overclock back to stock 2GHz. The system stayed at the POST screen for a long time, ostensibly testing the RAM, even though I had the quick memory testing option turned on in my BIOS settings. However, it was not showing me the amount of RAM tested in slowly ascending numbers as normal, but was instead stuck at the message: "Testing Memory: 4193". I have no idea if this number represented amount of RAM in MB, a memory address, or an error/test code. The number seems about right for amount in MB, but there was no "MB" after it.

I was not able to enter CMOS setup during this time. It wouldn't let me, it wouldn't respond. I therefore thought the computer had frozen again and I reset. Same thing happened again, but I decided to leave it. After a really long time (over 10'), the computer eventually booted into Windows!

Note that apart from throwing my CPU back to stock speed, it didn't reset any other CMOS settings back to stock. I know cause I turned off the green Gigabyte splash screen, and I'm still not getting it, it's just showing me the normal POST screen with the BIOS version etc.

I was not able to consistently reproduce the problem. I restarted 3 times. Twice it booted into Windows (once after a few minutes stuck at POST again, the next time lightning-swift as it should do). The third time it stuck at POST again, but this time it didn't even get to memory testing stage, I've just been stuck here staring at "Awawrd modular BIOS v. 6.00" for the past 5 minutes until I lost patience and reset. Again, it won't let me enter CMOS setup. The fourth time, it booted into Windows immediately!!!

I need some general advice on what's likely to be wrong, because the only thing I know how to do is to tear it down completely and rebuild it with minimal components, possibly swapping bits out until I find out what's wrong. However, I can't really do that, first because I'm really busy these days and don't have that kind of time, and secondly because of the strange inconsistency of the problem: it seems to be working just fine about half the time, so swapping components out isn't likely to help me find out which of them is faulty!!

Really frustrated and confused, and need OCUK's help!

Using an C2D E2180 (formerly at 3GHz), a Gigabyte P35-DS3, a Powercolour HD3850, an Antec NeoHD 500W and 4GB of Crucial Ballistix RAM.

 

[MooMoo444]

Sounds nastily like a dodgy motherboard - if its not letting you change CMOS, take the battery out and leave it a while.

Also try taking all but 1 stick of RAM out, and see how it goes. Could be a faulty stick, if it still runs bad with 1, then that may be the faulty one, so swap it for another and try again.

Doesn't hurt to try

 

[manveruppd]

Running bad is a relative term though, it hasn't even crashed once since I made that post. So whatever I do I might have to wait for days on end before I can safely say that this was what was causing the crashes!
I suppose I could spend an evening or two running memtest... but if my motherboard is dodgy, might that not cause memtest to produce errors even if the RAM is fine?

 

[manveruppd]

Got the "testing memory: 4193" thing where it just stays there for ages again, and this time it DID let me enter CMOS setup - it just did so a really long time after being stuck at memory testing for around 5' or so, so obviously I just wasn't patient enough with it last time and rebooted before it got unstuck... Tried upping the chipset voltage to see if it might help with this instability, rebooted again, and it wouldn't even get to POST, just a blank screen. So I'll be clearing the CMOS jumper later on, just posted to confirm that the overclock definitely had nothing to do with the error, cause it was still at stock when I got the freeze during "testing memory" again.