Trouble understanding the Memtest results and how I should act next

[KaelanFrey]

Hey guys. A month ago, I had a MEMORY_MANAGEMENT BSOD, I don't remember the specific code. Shortly after, the PSU broke and I was only able to get a new one now.

I only ran Windows for a few minutes, and I didn't have any BSODs, however I decided that it would probably be a good idea to run memtest to make sure the error doesn't happen again without knowing if it's coming or not.

I've been running this program for 3 hours now, and I'm planning to let it run overnight. Errors have been found already, but I'm not sure how to proceed should the error count halt or continue to increase. Here are the parameters:

CPU: AMD FX-6350 @ 3.9 GHz
L1 Cache: 80K 33518 MB/s
L2 Cache: 2048K 24951 MB/s
L3 Cache: 8192K 10212 MB/s
Memory: 4078M 5061 MB/s

Iterations: 3
AdrsMode: 64bit
Pass: 2
Errors: 27

Error Confidence Value: 222
Lowest Error Address: 00083bb6cf0 - 2107.7 MB
Highest Error Address: 00083bb6cf0 - 2107.7 MB
Bits in Error Mask: 00000040
Bits in Error - Total: 1 Min: 1 Max: 1 Avg: 1
Max Contiguous Errors: 1

Test 4: 11 errors
Test 7: 16 errors
The other tests report no errors.

--

Should I be concerned? This report doesn't seem very threatening like reports from different computers with really bad RAM, but I am not experienced enough in this subject to make a judgment.

EDIT: After leaving the test overnight, memtest has counted no more than the original 27 errors. Apparently, all of them had been found in the first 3 hours of testing.

So, I'm going to repeat this test all morning. I'll send updates this way when I can.

[orioon]

I would try to determine if the results differ, when the system was switched several hours off, compared to running for several hours before starting memtest.

Could be some kind of cold boot issue which is a bad voltage design/capacitors gone bad in most cases.

You should also try to configure the lower and upper memory limits in order to stress the area harder where the error was detected (2107.7MB).

[KaelanFrey]

I would try to determine if the results differ, when the system was switched several hours off, compared to running for several hours before starting memtest.

Could be some kind of cold boot issue which is a bad voltage design/capacitors gone bad in most cases.

You should also try to configure the lower and upper memory limits in order to stress the area harder where the error was detected (2107.7MB).

Hello orioon, thanks for the advice.

I have ran another test for about 5 hours and no errors were detected. Both tests (the overnight one and this one) had the default settings.

If I run into another MEMORY_MANAGEMENT BSOD in the near future or if I catch a string of them in the far future, I will run tests following your parameters. If by chance I don't find errors in the RAM then, what else could cause this BSOD?

Also, your "bad voltage design/capacitors gone bad" argument makes sense. The PSU I had was a very old LC-Power one, low-quality and brittle, which ran on my computer for 7~ years over 2 builds. However, I never had a MEMORY_MANAGEMENT BSOD until it happened, but it makes sense that the PSU could have part of the guilt since it died literally 3~ hours later under stress.

I'm glad it didn't damage any other components (I tested the PC very briefly with another PSU, but the fan on that one didn't work), I was actually afraid the RAM could be the one thing that got damaged - everything else worked.

[David (PassMark)]

The errors were probably real errors. But they might only manifest themselves from time to time when running in Windows. As suggested above it might be also temperature related or due to some other external factor (e.g. EMI).

Give that you only have 4GB of RAM (worth about $25) I would swap the RAM at the first sign of instability.

[orioon]

... LC-Power ...it died literally 3~ hours later under stress

Well, doesn't make me wonder.
Makes my theory more likely in this case I guess.

LC-Power PSUs are not able to deliver the wattage from the package.
All serious tests I know of caused them to burn down under full load.
Lucky for you that nothing else took damage.

Some PSUs are semi-passive, the fan might only turn on once a certain temperature has been reached. Which PSU are you using now?

There can be many causes of memory errors.
In Memtest4 there have been reports that network traffic can cause memory errors.
Let's speculate around if the error returns, no need for that right now