Memory failure few months after previous RMA

[agurenko]

I've had a memory problem previously described here. I've got a successful RMA even though it took more than a month to get the memory replaced, but I've got a brand new (Manufactured in May 2022) pair of memory sticks. I didn't run a full memtest86 upon reception of those sticks, but a first hour didn't show any errors, so I've proceeded and re-installed my OS and wiped my drives to get rid of the issues generated by previous memory modules.
Everything's been running perfectly fine from May 23rd (RMA received) until 2 days ago. Close the work day's end multiple applications simply started crashing with segfaults in system logs. Same night I left the memtest86 running and the results are simply shocking:





After another day yesterday, in addition to crashing apps, I see my filesystem error count also started to count up (while staying at zero for those almost 2 months).

While I'll be surely contacting the memory vendor again, this is a quite baffling situation. How can a brand new set of memory got "damaged/corrupted" in less than two months? Does it indicate the problem with Motherboard/CPU?

I'm not running any overclocking, rather than XMP profile enabled, CPU is running at stock, no PBO enabled.

Any help or advise is appreciated.

[David (PassMark)]

It does look like a memory fault. (small address range and small number of bits in error).

So maybe just really unlucky. Or maybe something is electrically damaging the RAM. (Power spikes, static electricity, corrosion on the pins or sockets).

[agurenko]

It does look like a memory fault. (small address range and small number of bits in error).

So maybe just really unlucky. Or maybe something is electrically damaging the RAM. (Power spikes, static electricity, corrosion on the pins or sockets).

The pins look fine from a visual inspection, same for the modules. I've swapped the modules and the error count went a little down to ~2800 and the system is a bit more stable, but the weird part is the memory error address didn't change, which IIUC should've been shifted as it starts from a different module, so maybe it's the slot issue? I cannot try different slot combination in dual-channel mode as my CPU cooler blocks the closes to CPU memory slot

[David (PassMark)]

I've swapped the modules ....... the memory error address didn't change,

Memory addresses are interleaved in dual channel mode (in a complex fashion). So swapping the order of the modules won't change the faulty memory address much at all.

Can you borrow a known good set of RAM to test. Or test 1 stick at a time.