Desperately needing help with RAM test results...

[SalPancho]

Hi! Hoping I can get some feedback on this frustrating issue I've been having...

I started getting BSODs (stop errors) during heavy resource programs (games mostly). After testing my RAM with MemDoc, Memtest, and WinDiag, I've gotten a dizzying variety of results.

I have not been able to duplicate the errors in any consistant way. Sometimes, I will run it for 8 hours without a single error. Other, I'll get 20 errors on the first pass. Even other times, I won't get an error for 20, 30, 40 passes. The problem is the frequency of the errors, and the number of errors seems completely random. WinDiag identifies the errors in banks 2/3 90% of the time, and 4/5 the other 10%. Never are there errors in both slots at once...the errors will only appear in one slot. After switching out PSUs, I've ruled that out as the issue.

I've also tested all for sticks of ram one at a time in the 2/3 slot. They all passed 12 of the extended test runs in WinDiag. The errors really seem to happen when all the sticks are installed at once, but the configuration does not seem to matter.

My specs are below. Am I safe to assume that this is a Mobo issue? I've updated the BIOS, etc...I understand that the Mobo may not be faulty so much as not like the ram.

EVGA NF47 Mobo
AMD 64 X2 4800+
4x512 Corsair XMS PC3200 (2-3-3-6, but I've tried other timings with no luck)
EVGA GeForce 7900GT

 

[JMecc]

Although I've never seen test results that inconsistent, the common addage is that ANY error found by memtest means bad memory. This indicates a spot that will not repeatably spit back out the same value it is given, which can definitely crash a program using that spot. Not every time, not repeatably, but if there is an error found you are running the risk of getting bluescreened anytime. Return the memory where you got it.

Jo

 

[SalPancho]

Although I've never seen test results that inconsistent, the common addage is that ANY error found by memtest means bad memory. This indicates a spot that will not repeatably spit back out the same value it is given, which can definitely crash a program using that spot. Not every time, not repeatably, but if there is an error found you are running the risk of getting bluescreened anytime. Return the memory where you got it.

Jo

Well, I got the all 4 sticks about a year ago. Besides, I am not able to determine which stick it is, as one stick failing in banks 2/3 (slot 3) will pass in banks 0/1 (slot 1). Is it not possible that the memory slot on the mobo is bad?

 

[JMecc]

It definitely is possible that the motherboard has an issue. Try with ram you know works (passes tests on other machines). Also, test only 1 stick in bank 1 & if that passes, test the same stick in the other banks to see if a bank is bad. If they all work with that stick, the banks are good, so try each stick by itself in bank 1.

Jo

 

[SalPancho]

It definitely is possible that the motherboard has an issue. Try with ram you know works (passes tests on other machines). Also, test only 1 stick in bank 1 & if that passes, test the same stick in the other banks to see if a bank is bad. If they all work with that stick, the banks are good, so try each stick by itself in bank 1.

Jo

I have tried that, and the results are very inconsistent.

Failures ALWAYS happen in bank 2/3 (slot 2). On 2 occassions there were errors in 4/5, but NEVER in the other 2 slots regardless of which RAM stick is in it. When testing the ram one by one, I tested them all in slot 2 (which all of the errors have appeared in) and they all passed through 12 cycles of tests with WinDiag.

The inconsistancy is very frustrating, as pinpointing the root is a crap shoot. I've already ordered a new Mobo, as I would figure that if it was a single RAM stick it would fail consistantly...

 

[Mondoman]

SP, it's not clear if you used memtest86+. If not, go ahead and run that.
If it happens during heavy resource use, one or another component may be overheating. Please report CPU, GPU, and any other temps during Windows idle, and during heavy use. A good program that will stress the CPU and RAM (but not the graphics) is Orthos (run under CPU and memory stress mix).
Also, are you overclocking at all?

 

[SalPancho]

I have used Memtest and that's were I first found errors. However, since WinDiag can identify the slot the errors occur in I switched to using that to determine exactly what was wrong. Memtest did not find errors on the first pass...it would not find anything wrong until 20+ passes.

I am not overclocking, and my CPU/SYS temps are around 32-35. My GPU is around 45 idle. I'll have to check the temps under load later, but I've never had other overheating issues (artifacts/tearing, etc.) and I don't have any abnormalities or crashes outside of games (which are pretty much all I run on the PC).