Intel, responsible for widespread use of non-ECC memory, as well as enforcing a new socket every one or two CPU generations.
Infamous for recent two consecutive generations of self-destroying CPUs (13th and 14th gen).
What a load of bull.
Infamous for recent two consecutive generations of self-destroying CPUs (13th and 14th gen).
What a load of bull.
But in general it speaks to the anti-consumer behavior of Intel.
I will just highlight that Linus Torvalds disagrees[0], and that this sort of thing[1] happens whenever there's a geomagnetic storm.
0. https://arstechnica.com/gadgets/2021/01/linus-torvalds-blame...
It's especially not worth it if you're storing all your critical data on cloud services that already run on server-grade hardware.
Why do I care about ECC memory if the source of truth of my data is in cloud services like S3 that have impossibly high durability standards?
We can even back it up a bit and say, how do you know that data loss isn't acceptable to me?
Is ECC memory really more protective compared to a 3-2-1 backup system? ECC memory can't stop my house from burning down or having water intrusion destroy my system. I think a lot of those "natural" causes of data loss are more statistically probable.
Finally, it's my understanding that filesystems with checksums can essentially mitigate this problem or warn you very early of issues.
100% irrelevant, sure.