not to step on any toes,, but,, 3 passes on 12G of memory won't even begin to stress it.
You need to run your tests for no less than 8 hours on that much memory.
However,, for a quick test,, yes, a few passes (more than 3) should suffice, but you won't really know till you run a stress test of at least 8 hours on that much memory. If you do get errors within the few passes then you either have a bad stick or a slot
and you will need to test all the sticks and slots individually, till you find the bad one. If the error pops within a few passes, then you don't need to test for that long on each one,, just a few passes till you find the bad one.
I would have to look around to find the reason for the long stress tests to give a full explanation as to why you want to run them that long, but the gist is,, it takes that long to test every part of the memory. More so for intermittent problems. There are some deep technical reasons that I just don't want to try and find again. They say you get 100% coverage in a single pass,, that's only partially true, to what I remember.
Similar to running a surface scan of an HDD.
I forget which company I was talking to for hardware diags,, might have been PC Doctor or Ultra-X or both. But they confirmed the need to run long overnight memory tests to test all bits of memory fully.
If you have bad memory and you fill it up, you will get BSOD's, crashes and/or data corruption, at random.
If you run it for a long time,, 8+ hours,, not to exceed a certain amount of hours,, 12G is a lot to test... I generally say 4 to 6 hours on 4G of ram, then add about a half to an hour per 1 to 2 gigs,,, it's not a perfect science. No hard rules,, just a generalization.
This might help
Also keep in mind that Memtest86 or memtest 86+ do throw false negative and positive errors. To be sure,, you should also run Windows memory Diags,, I have actually found that Windows memory Diags do a better job and throw far less false +/- errors.
edit:
hehe,, I got caught editing my post,,