Wondering if this was broken in an agesa version and just never fixed.
I have some ecc micron udimms, and I suspect the ecc it’s self is working, but I can’t get anything to report errors. The system goes from very stable to unable to reach the desktop without freezing with only minor timing adjustments and some very aggressive timings report no errors in memtest.
ECC reports as on and working as well, but none of the corrected errors show up with edac-util or ras-mc-ctl, or in dmesg. Journalctl also doesn’t seem to show anything.
x570 doesnt officially support ecc. any support there is, is a patch on solution.
there is nothing on the cpu to tell the ecc mem controller if the ram is erroring.
so in effect the ecc chip on the ram has no controller so cant confirm the error was fixed or even if it was an error to begin with.
1 Like
I’ve seen success with other boards reporting ECC to the OS, in various topics on reddit or even on L1 Forums, so there’s definitely something that can feed that back in to the OS, but it seems kind of flaky to get it to play nice at times.
My understanding has been that AMD has ECC UDIMM support in the memory controller/CPU fully, same as their pro lineup, but that it hasn’t been tested/validated in order to reduce cost of authoring consumer AGESA revisions, and therefore isn’t a listed feature, similarly to memory support for more than 64GB on 1000/2000 Ryzen CPUs.
And wow is ASUS support tree completely broken. Some of their lines of contact just end in a blank page with no contact information at all, and there doesn’t seem to be any email?
I also know that the Prime X570-p does work and is able to report to the OS, from other topics on the L1 forum.
The problem is, I don’t know which agesa, kernel, or util.