Run MemTest >3 days?

[LeoBloom]

I will be away from home for 4 days - is it OK to run MemTest while I am away for this long?

I won't have any remote access or somebody from home that could turn it off

I have never run it beyond 24 hours, but lately have been having BSODs, but need daily access to computer (apart from these upcoming 4 days) so I can't run multiple passes

[David (PassMark)]

By default MemTest86 will stop after 4 passes (typically a couple of hours).
If it runs 24h without error, the chance are low that real memory errors will be detected beyond this period.

Eventually if you test long enough you'll get a soft error however.

[LeoBloom]

Thank you for this answer

[rock eaters]

Hi David,

we have the same problem as described. After some hours, spantanoues one of the 13 test fails. It is fully happenstance which one. Sometimes the one and if we restart the test, after the 72 hours or more an other test fail.

Did you can explain that?

Best Regards
stone eaters

[David (PassMark)]

We'll clearly the RAM isn't reliable over a long period. To some degree this is just the nature of DRAM, and the reason error correcting ECC RAM was created. See my comment about soft errors above. But you should be able to get an error free period longer than "some hours". I would be tempted to go back to the vendor and ask for the detailed specs for the RAM you are using and find out what error rates are expected in long term use. If you see rates higher than the specs, then it is faulty.

[rock eaters]

Hi David,

can you guarantee, that the software didn't have a problem with multi-threading?

Best Regards

[David (PassMark)]

We have 1000s of people using the software daily. We aren't aware of any problems with the software.

We are aware of firmware bugs in some UEFI motherboards however. But these generally cause a freeze of the machine, not false positives.

[rock eaters]

Hi David,

I thought I had the problems, if I use all CPUs in parallel(second picture), but now I found out, that if I use "round robin" I get the problem too.

But it need a lot of time. Mostly 48 hours or more. If it should be soft error, than i would expect, that the problem comes not so often, and need much more time, and that only one address is affected and not the complete Memory. But in all my tests the complete memory is suddenly corrupt. But nothing more then the tool is running and I did no changes in the system during running.

The System even didn't freeze, but the tool shows errors all the time. The system itself run. So I think the memory, in which Memtest itself runs, seems to be ok...

Can I support you in some way? Log Files or somthing more??

Best Regards

[David (PassMark)]

The real test is to swap the RAM and see if the problem goes away.

[rock eaters]

Hi David,

I tried a lot of RAMs in i-temp or c-temp range, from samsung, micron and other.

But the Problem is always the same. First all the tests pass, 10 times, and after two or three days one of the thirteen tests start to fail.
I made an other test, where I used the round robin setting and even after 49 hours the test start to fail.

Any advise or hint?

Best Regards

[rock eaters]

here is a part from the log -files:
2019-01-03 02:56:20 - RunMemoryRangeTest - CPU #1 completed but did not signal (test time = 319013ms, event wait time = 1090ms, result = Success) (BSP test time = 321416ms)
2019-01-03 03:34:33 - RunMemoryRangeTest - CPU #2 completed but did not signal (test time = 570310ms, event wait time = 1001ms, result = Success) (BSP test time = 575951ms)
2019-01-03 03:34:33 - RunMemoryRangeTest - CPU #3 completed but did not signal (test time = 570338ms, event wait time = 1044ms, result = Success) (BSP test time = 575951ms)
2019-01-03 03:34:33 - RunMemoryRangeTest - CPU #1 completed but did not signal (test time = 572365ms, event wait time = 1090ms, result = Success) (BSP test time = 575951ms)
2019-01-03 03:34:34 - RunMemoryRangeTest - Could not start AP#1 0x000000006B020000 - 0x000000006D23BC00 (Time out). Resetting...
2019-01-03 03:34:34 - RunMemoryRangeTest - Unable to start AP#1 0x000000006B020000 - 0x000000006D23BC00 (Time out)
2019-01-03 03:34:34 - RunMemoryRangeTest - Could not start AP#2 0x000000006B020000 - 0x000000006D23BC00 (Time out). Resetting...
2019-01-03 03:34:34 - RunMemoryRangeTest - Unable to start AP#2 0x000000006B020000 - 0x000000006D23BC00 (Time out)
2019-01-03 03:34:34 - RunMemoryRangeTest - Could not start AP#3 0x000000006B020000 - 0x000000006D23BC00 (Time out). Resetting...
2019-01-03 03:34:34 - RunMemoryRangeTest - Unable to start AP#3 0x000000006B020000 - 0x000000006D23BC00 (Time out)
2019-01-04 23:51:09 - Attempting to retrieve the root directory of the data partition.
2019-01-04 23:51:09 - Found 2 handles that supported SimpleFileSystem
2019-01-04 23:51:09 - Checking handle 0 (752549D
2019-01-04 23:51:09 - Handle is own partition
2019-01-04 23:51:09 - Checking handle 1 (7524ED9
2019-01-04 23:51:09 - HD node found: MBR type(2), Signature type(2), Part #(2), Start LBA(51404, Num Sectors(509919)
2019-01-04 23:51:09 - Self node: MBR type(2), Signature type(2), Part #(1), Start LBA(204, Num Sectors(509953)
2019-01-04 23:51:10 - Disabling watchdog timer (Result: Success)
2019-01-04 23:51:10 - Could not find Simple Network protocol.
2019-01-04 23:51:10 - =============================================
2019-01-04 23:51:10 - MemTest86 V8.1 Pro Build: 1000 (64-bit)
2019-01-04 23:51:10 - =============================================

[keith]

It appears to be an issue with the multiprocessor subsystem of the UEFI firmware. Can you check to see if there is an updated BIOS.

Also, another test to try would be running in single CPU mode and comparing the results.