memtest86+ error explosion, and then nothing

Question

Over the last two weeks my desktop has seemingly all of a sudden become noticeably less stable: frequent BSODs (all with different error messages but all of which were related to memory), difficulties performing basic application installs/uninstalls, random application misbehavior, and occasional "forgetting" to load drivers at boot-up (for instance, on one boot, windows wouldn't load the graphics drivers and the resolution would be terrible; reboot, everything's fine again. wash, rinse, repeat).

The only common thread I could think of was RAM, so I booted into memtest86+. Within minutes it had found thousands of errors; over the course of the night, it found 125,000+ errors in 4 passes.

So I started swapping them out and rotating the DIMMs. I have four sticks of memory for four slots, each 2 GB. Starting with the stick in the first slot, I tested it by itself in slots 1, 2, 3, and 4, going for at least two passes in memtest before moving it to the next slot. No errors reported anywhere.

Next, I tried the stick that had originally been in slot 2 and put it back in slot 2 by itself. No errors. Same with stick 3 into slot 3 by itself. No errors. Same with stick 4. No errors.

Then I tried sticks 3 & 4 in the last two slots. Nothing. Sticks 2, 3, & 4 in the last three slots. No errors. I am now running all four again - the exact configuration that produced the cascading memory errors the first time - and so far it's been running for almost 30 minutes (well into the thousands of errors from the previous scan), and it hasn't reported a single error.

Is my computer just possessed?

(setup: Intel Core 2 Quad Q9300 ; 8 GB Corsair Dominator ; eVGA 750i SLI motherboard ; 1x nVidia GeForce 8600 ; 680W PSU)

Answer 1

It is possible that the memory seating had problems for a while.
You activity with the memory modules get it working again.
I have seen this happen (in exactly the way you describe).

Additionally, there was also a case when the PC would just not boot and declared memory failures. Removing the replacing the memories made it work fine (and completely pass the memtest).

^{I wonder these days if part of the PC cleanup activity should also include cleaning the card (memory and other PCI) contacts... suggestions on this are welcome.}

Answer 2

This sort of behavior can also be caused by a bad power supply or CPU (or even an overheating one that's not actually bad). It could also be a bad motherboard.

If the problems go away after reseating the RAM, than that's well and good.

I was not so lucky, in my case it turned out to be an unstable power supply.

Answer 3

Reseating memory is sometimes known to fix errors (sometimes, depending on the cause of the errors). To learn more, you can research "reseat RAM oxidize", and that may help to make the cause seem a bit more scientific rather than mysteriously random.