7.5: Always the same single one bit error. 4.3.7: no error at all.

[Couac]

Hello,

I have run 3 different tests using memtest 7.5. During those tests memtest showed me the exact same result: one single bit error during the same hammer test at the same address. Also, only the first test seems to fail, all subsequent tests all seem to pass (on my third 7hours test, 9/10 tests passed)
Also a 5 hours memtest 4.3.7 showed no error at all.

Is my memkit defective?

Thank you a lot!

[David (PassMark)]

Errors in the hammer test are special. Not as serious as normal RAM errors. See this page for details.

V4 of the software doesn't have a hammer test. So no surprise it didn't detect an error.

[Couac]

Thank you for your quick answer. If I understand things correctly, those errors are deemed non-critical and even if I were to RMA my memkit the manufacturing company would probably consider those errors as normal? I ask this question because I just bought this memkit, do you think I should try an RMA? From what I understand the only way to get rid of this kind of error would be to try another model/brand, am I right?

[David (PassMark)]

Depends what you use the PC for. It it was a home machine just used for gaming then I would live with the problem. Likely you won't see the problem very often in real use and the consequences of failure aren't dramatic. It if was safety critical system (medical equipment, air traffic control system, mars probe, etc..) I would definitely change the RAM.

GSkill offer unlimited warranty
https://www.gskill.com/en/warranty
So if it gets worse in a couple of years you can still change it.

[Couac]

How often do you think I may encounter problems like BSODs? My computer is a home machine but I still use it for work from time to time. A crash once or twice a year would be bearable but once every month would be a problem for me.

If I understand what you said correctly you still think my memkit is defective? Reading the article you linked I understood that replacing my memkit by another of the same model would probably wield the same result, was I wrong? Thank you a lot for your help.

[David (PassMark)]

How often do you think I may encounter problems like BSODs?

It is semi-random. Impossible to say. Could be never. Could be immediate after each boot (although probability of this is low).

you still think my memkit is defective?

Yes. But we have a similar machine in the office. Throws one hammer error. It goes for 6 months in real life use without a problem. It is random however.

I understood that replacing my memkit by another of the same model would probably wield the same result, was I wrong?

Maybe, as it is more a design flaw than a manufacturing flaw. But as you get just one error, it might be a combination of both. One slightly weak cell in RAM that doesn't have a lot of tolerance for weak cells.

[Couac]

Thank you a lot. I will try and initiate an RMA with the store I bought it from.

[Couac]

I really don't understand what is happening. Because I was told by newegg support that if they didn't find any error on their end they would send me back my memkit, I decided to run another overnight memtest. It didn't find any error (test4.png). At all. Then I've run another test, selecting only the row hammer test, only testing the memory range impacted the first times. Same, no errors at all (memtest5.png).

I did not do anything. Well I did reduce the DRAM Refresh Interval in the BIOS, I have then launched a memtest, which did not find any error so I put the DRAM Refresh Interval back to auto, which did go back to the value it was before I messed with it, and then launched the memtest pictured in test4.png, but I highly doubt it has anything to do with the memkit now passing the row hammer test.

Please, do you have an opinion on the matter? Thank you.

[David (PassMark)]

Fixed itself. Magic.
Problem solved

[Couac]

errm I was hoping for a more enlightened answer, if the problem disappeared on its own it may come back on its own, but anyway, thank you a lot for everything

[David (PassMark)]

Arthur C Clarke:
"Any sufficiently advanced technology is indistinguishable from magic".

In short I don't know and it is unlikely anyone else will know either. Row hammer errors are basically EMI events. So either the EMI environment was different. Or the memory cell really did improve it's performance, or it's performance is influenced by some external condition. e.g. temperature. None of which are going to be diagnosed remotely from a forum post.

[Couac]

Arthur C Clarke:
None of which are going to be diagnosed remotely from a forum post.

...or by a newegg technician who is testing his 10th memkit of the day.

Anyway, thank you for taking the time to explain what is wrong with my memkit.