The plague of weird issues strikes again...

[Ericloewe]

The G-550 has been RMA'd... And replaced with one of the new units, supposedly with a crap Hong Hua glorified sleeve bearing ("FDB") fan instead of the old Yate Loon Double Ball Bearing.

Fun little detail I just noticed (after struggling to get memtest86 to boot in UEFI mode): The X11SSM-F has such an old BIOS/BMC/ME firmware release that the X10SLM+-F actually has a newer version of the Java IPMI viewer:

X10SLM+-F left, X11SSM-F right.

And the X11 BIOS is easily the buggiest BIOS I've had the dubious honor of dealing with. Not worst (there was this old Phoenix BIOS that started beeping the speaker, which was still a real speaker and not a tiny piezo thingy, whenever the sensors (temperature and fans) page was accessed), but easily the buggiest.

 

[gpsguy]

btw, today I read that MemTest86 7.0 was released - http://betanews.com/2016/07/21/memtest86-7-0-now-faster-detects-more-ram-errors/

Fun little detail I just noticed (after struggling to get memtest86 to boot in UEFI mode),,,

 

[Ericloewe]

btw, today I read that MemTest86 7.0 was released - http://betanews.com/2016/07/21/memtest86-7-0-now-faster-detects-more-ram-errors/

Odd timing. The beta was next on my list, mostly because it's actually the pro version that supports ECC error injection.

I'm running it with memory scrambling on and I was going try it afterwards with that option off. The idea is mostly to see what happens...

 

[Dice]

The X11SSM-F has such an old BIOS/BMC/ME firmware release that the X10SLM+-F actually has a newer version of the Java IPMI viewer:

>.<
that ...
:(

 

[Ericloewe]

And, just like that, the DIMM is no longer throwing occasional ECC errors.

Once the current batch of tests is done, I'll run prime95 again for a few days to soak in the heat and then try memtest again.

If the ECC errors do pop up again, I'll just RMA it, since it's neither 100% stable nor unstable enough to be of use testing ECC.

Also, the memtest86 7.0 Beta (which is the pro version) fails to inject ECC errors on the system. It'd be interesting to see the results on a Xeon E3 v3/v4.