Bad RAM? Help Interpret Memtest86 (28,416 errors then 0 errors)

I'm really confused on whether I have bad RAM or not because of the memtest results I'm getting. I've already had to replace the mobo and graphics card in my nearly brand new built desktop as I had problems with both (mobo wasn't W7 compatible, graphics card crapped out).

As I had two major problems, I wanted to make sure my RAM was 100% okay. I currently have four 2GB Corsair Dominator DDR2 sticks of RAM (two matched pairs). Here is what I came up with via memtest...

The first time I ran the test, it came back with 28,416 errors in the first pass alone. Here is a screenshot.


I'm not sure what all the details mean other than there are a LOT of errors there. Any help interpreting those results?

So I knew I have potentially bad RAM or a bad slot on the mobo. I pulled two of the sticks (one of the matched pairs), then retested via memtest. I let all these memtests go for seven passes and 10 hours of run time just to be 100% for sure. 0 errors with those two sticks pulled.

So then I wanted to verify I didn't have a bad slot, so I moved the two known good sticks to the two now-empty slots. Ran memtest, 0 errors.

So then I knew at least one of the two sticks I originally pulled must be bad. I plugged one stick back in and ran memtest again to see if it was the bad stick, 0 errors.

So now I think I have the bad stick isolated to the only one not in the board now. But I want to verify and be 100% for sure, so I plug it in and ran memtest again fully expecting to see a ton of errors again. It's been seven passes and 10 hours now and 0 errors.

What gives?

This is the same four sticks that gave me 28,416 errors, with the only difference being the fact that they moved around in the slots due to my memtesting, and now they're giving me 0 errors. How is that possible? Do you guys think I do have bad RAM? If all my RAM is good, why would the first time I ran the test report a whopping 28k+ errors?

Thanks a ton for any help here. I'm FAR from tech-savvy and I'm learning all of this as I go so I appreciate all the help I can get.

Well, there is the possibility that some clutter got into the ram slots when you put the ram in the first time, which cleared out when you reseated the ram.

The other possibility is that the ram wasn't properly seated until you reseated them.

I was thinking one may not have been completely seated the first time around with all the errors. As they're all in there now and showing 0 errors I should be good to go, correct?

You should be fine if you ran the tests several times and didn't get any errors. Of course, Memtest86 is only good for the basic bad memory testing, Windows will definitely put the memory through much more rigorous testing based on actual speed timing and voltages as another thread showed for Jalepi as far as stability is concerned.

That appears to be the case. While memtest checked out fine, I've had several BSOD crashes that are directly related to memory errors yesterday when I was really taxing the system. Looks like I've got faulty RAM somewhere. For now since memtest checked out okay on all sticks, I've pulled two and I'm hoping for some stability for a week or two, then I'll contact Corsair about replacing the sticks I pulled if it turns out they're causing the problems.


Quick question related though. I did some searching in relation to my motherboard to see what RAM Gigabyte recommends, and they say 2x2GB DDR2 PC2-6400 1.8V. Since the Corsair I have is PC2-8500 2.1V and Gigabyte recommends PC2-6400 1.8V, would that make any differences or cause any problems?

I'm a complete rookie on this sort of stuff, so I'm not really sure how to set them to 1.8V. I do know the RAM was overclocked in the BIOS initially (set to turbo) straight out of the box, and it was causing massive crashes left and right. Now it's only when things get really intensive that it crashes (Lightroom exporting big RAW photography files). How can I verify the actual voltage and set it to 1.8V if it isn't already?

Okay, I'll poke around the BIOS and see if I can find something.

Gigabyte, per their website, actually says the ram is part of the QVL list and compatible. Corsair CM2X2048-8500C5D is my RAM (I have two pairs of the 2x2GB sticks), Gigabyte GA-EP43-UD3L is my mobo. They're compatible. I was just wondering if the voltage difference could be causing an issue (I've heard it both ways, run your RAM at 2.1 regardless of the mobo voltage, or only run it at mobo voltage regardless of RAM voltage). I was just hoping for someone with real tech knowledge to say what the real right way to do it is.

Tews has plenty of tech knowledge for this. You need to run the memory at manufacture specs. Manufacture being Corsair. So run it at 2.1 volts and check the cas latencys also. Your board should handle it either way and should have a way to adjust the memory voltage.