Why is memtest repeating single test?

0

I've been having some problems with my computer. I'll start with a short history.

Bought it 11 months ago, immediately there were problems. What would happen is, when I logged in after a long period of shut down, the system would BSOD. Then I would restart and everything would be fine, until the next time I shut down for a long time (8ish hours) and got back on.

I sent the laptop in (a couple times) and they eventually replaced the RAM and the BSODs went away. This is all on Windows 7 (well started on Vista, switched to Windows 7).

I recently set up a dual boot with Windows 7 and Ubuntu 10.04. I'm experiencing crashes on Ubuntu (the whole system just freezes, can't operate mouse, keyboard, or anything) and the behavior is similar. When I log on after a long period of shut down, the system will crash, but if I restart (actually I have to wait 10 minutes before I restart, if I restart immediately it will crash again) then it is fine, until the next period of long shutdown.

So I'm thinking it's bad RAM, so I downloaded Memtest and am testing each of my sticks (they are each 2GB) individually in the same connector thing (don't know what it is called, whatever the sticks attach to). The test for the first one passed with flying colors. However, the second one is having problems. On test #2, the memtest repeats itself, continuously looping through 0-32767 errors on the right side.

Does this indicate that the memory is bad, or is it a problem with memtest? Any help is appreciated, thanks.

P.S. My laptop is an Asus G50VT

Edit:

The 2 most interesting lines...

Highest Error Address: 000fffffffc - 4095.9 MB // does this make sense? 4095.9 MB on a 2GB stick?

Test 2: 32767 // the 32767 loops back around to 0 and repeats

random

Posted 2010-06-20T23:11:02.687

Reputation: 3

Answers

0

First, I recommend using memtest86+ (4.1 is recent) if you aren't already. Sadly the version matters, and newer ones are more compatible with whatever hardware IME.

Second, you mention that you have two sticks. If you are testing them at the same time, then 4095.9 makes sense. If you are testing them one at a time, it doesn't, and I'd suggest trying a different download of memtest. If you're using the newest memtest86+ and it also reports byte 4095.9M, it's conceivable that the problem is with something else, like the CPU or motherboard. Still worth trying to replace the RAM though.

Third, if you have so many errors that you overflow the counter, that's what's going to happen. It will go up to a power of two, and then go back to 0.

Slartibartfast

Posted 2010-06-20T23:11:02.687

Reputation: 6 899

I'm only testing one at a time. Why would the other stick work fine though? Of course unless this is an indicator that the stick currently being tested is bad. That's the thing, I have no way of knowing if it is the stick or memtest, but the fact that the other stick was fine leads me to believe that it might be the stick. – random – 2010-06-20T23:27:34.147

I've seen some strange errors with memtest. I can't explain them. (I don't know why one would be fine and the other have issues) – Slartibartfast – 2010-06-20T23:29:03.270

I shut down and booted it up again. Seems to be working fine now. Will let you know. – random – 2010-06-20T23:32:23.073

Test passed, but I read about a Test #9, and I don't think my version had that. So going to update to memtest86+ and run through them both again. – random – 2010-06-20T23:57:56.490

Overflow alone doesn't explain this, as overflowing from 32767 to 0 would be a 15-bit unsigned int. It is suspicious of overflow, though. – None – 2010-06-21T02:40:05.147

It is overflow for a 16-bit signed integer + a little cautious coding to prevent negative numbers. If I were the author I'd say: "Negative numbers make me look stupid. >30k errors make the hardware maker look stupid. I'd rather the hardware maker look stupider than 30k, but I don't care that much." – Slartibartfast – 2010-06-21T03:08:50.533