ExedExes wrote:That's news to me about RAM passing POST, in all my years it's never happened to me. I do have Memtest x86 already burned to a CD in case it's needed, it's part of my CD toolbox, alongside other stuff like the Hirens CD which is a good live CD as long as we're mentioning those.
This thread is way past the original purpose but I appreciate a place to talk about what I do on a daily basis.

I was "lucky" enough to have the strangest PC problems in the past. They taught me more about diagnosing hardware issues than classes I attended at a trade school.
But yeah, I mean, that's why MemTest exists. If POSTing meant 100% good RAM, then why run a test?
Think of it like bit rot. How much of the 4+ gigs of RAM is needed to post and load the OS? Some area of RAM might be bad but you wont know until something is loaded into that area. For an analogy, lets say a CD/DVD has a small but deep scratch. The seller on eBay tested the game, got through the menus (let's compare that to POST) and loaded the first level (let's compare that to getting to the desktop). You buy the game, but when you get to level 5 the game hangs because that scratch contained data that wasn't needed until then.
That's why it's a good idea to run MemTest for as long as possible. And, I would assume, the more RAM you have the longer you should let it run. Luckily in that example I gave, MemTest returned multiple errors straight away. But could you imagine having 24GB of RAM and only a small area of one stick is bad? And maybe that one small area of RAM works 9 out of 10 times. By random chance you might never know, or by random chance you would only get an error or BSOD once in a blue moon that you would just chalk it up to "shit happens" and forget about it. As I understand it, MemTest writes patterns to RAM and then reads them back and checks if they match. How large can a pattern possibly be? How long would it take to do just ONE pass through 24GB of RAM? (I have no idea).
RAM can also be very tricky. When using MemTest, apparently some patterns can pass but one particular pattern will fail. Think of the random chance it can be of that happening while the OS is loaded and the PC is being used normally.
Says Wikipedia (I haven't used MemTest in a long time), there's another test that doesn't run by default. It writes all 0's to RAM, sleeps for 90 minutes, then checks the RAM to make sure none of the bits have changed. Then it repeats with all 1's. I remember in the 90's (people probably did this in the 80's too, but I was only born in 87) if your computer was doing wacky shit you would shut it off and count 30 seconds before turning it back on. The thought was that maybe something got screwed up in RAM (written wrong, or corrupted) and cutting power for 30 seconds should clear the RAM.
What I'm getting at is...
Xeogred wrote:Memtest also passed with flying colors after only about an hour of scanning.
You can run MemTest for an hour (or more) without any failures, but that doesn't mean your RAM isn't bad.
I have no reason to believe that RAM is definitely your problem, I'm just yapping here.
