got non-repeatable memtest errors

[ElFenix]

should i RMA? will probably do so, just wondering if non-repeatable error could be something else.

first test errored nearly immediately. subsequently run tests with both memtest and windows memtester have resulted in 0 errors.

everything's running at box stock right now.


background: getting seemingly random crashes in windows with new setup (3570k, asrock extreme4, crucial sport 1600), trying to figure out the issue.

 

[smirk]

What's the longest you've run memtest without getting any errors? Did you get the original error while running at stock speeds and timings?

I would RMA it, you shouldn't ever get memory errors.

 

[Lotheron]

seemingly random crashs + ANY memtest errors = RMA.

It could be very random or heat related.. but the fact that you're getting crashes and you've had a failed test means it's most likely bad.

Run it overnight and see what happens.

 

[Aikouka]

Which Memtest are you running: Memtest86 or Memtest86+?

 

[ElFenix]

Which Memtest are you running: Memtest86 or Memtest86+?

memtest86+. the first time i ran it, 1 pass only, it errored within a few mintues. a subsequent single pass didn't find any errors. a pass took about an hour and a half. overnight i let windows' built in memory test run (it had previously found errors that long memtest runs hadn't found on other computers), and it found no errors as of this morning. it's currently sitting at home running memtest86+, so will be about 8 or 9 hours by the time i get home.

i've not tried tweaking the computer at all yet, so all crashes have been at stock settings.

 

[Aikouka]

Interesting. I don't think I've ever had the Windows built-in memory test perform better than Memtest86+.

Have you considered trying it with different hardware (a different computer)?

EDIT:

Also, sometimes I swap around memory configurations. Typically, if I have say... 4 DIMMs in, I will then try and figure out which ones are actually bad. You can typically guess depending on where it screwed up (the address), but I'll usually do one at a time.

To then avoid the uncommon issue of a motherboard problem, I'll typically also shift the DIMM around to a different slot. Another PC like I mentioned earlier is also nice. I need to consider building a more dedicated test bed sometime. Right now, I just use my extra PC to perform tests.

 

[ElFenix]

Interesting. I don't think I've ever had the Windows built-in memory test perform better than Memtest86+.

Have you considered trying it with different hardware (a different computer)?

EDIT:

Also, sometimes I swap around memory configurations. Typically, if I have say... 4 DIMMs in, I will then try and figure out which ones are actually bad. You can typically guess depending on where it screwed up (the address), but I'll usually do one at a time.

To then avoid the uncommon issue of a motherboard problem, I'll typically also shift the DIMM around to a different slot. Another PC like I mentioned earlier is also nice. I need to consider building a more dedicated test bed sometime. Right now, I just use my extra PC to perform tests.

unfortunately this is the only ddr3 computer i have so can't parts around like i could ddr2.

isolating and testing slots will be in the future, but it sure would be nice if i could get some repeatable errors first.



i may have been using memtest86 previously.

 

[bryanl]

Memory with a testosterone-inspired decor or name should always be a prime suspect, but occasionally motherboard manufacturers issue BIOS updates to improve memory compatibility (slow some timings). Heat is never a factor with proper memory.

Microsoft's memory diagnostic, included with newer Windows, occasionally finds defects that Memtest86+ and Memtest86 miss, but that's rare.