I suppose people are used to crashing, glitchy flakey, crap that passes for PC's these days... *sigh*
You mean, you really saw even once in your life ECC memory to save your desktop or workstation from crash?
I don't think ECC makes any sense, at least unless system is constantly under extreme stress over 24/7 and that in very dense electrosmog.
The saying goes that there are only two kinds of people, those that use ECC memory and those that never had their RAID array written full of crap due to a faulty memory chip.
The subject you are talking about applies to workstation and desktop/ or server?
Have you already experienced the situation you describe?
I have 2x harddrives on my machine, but I dont do raid, but rather normal backups exactly for the reason that RAID is only partial backup. The situation you describe never ever happend to me and I can only imagine bit-flipping on very high-electrosmog devices such as rack servers. Can you report the opposite?
The situation you describe never ever happend to me and I can only imagine bit-flipping on very high-electrosmog devices such as rack servers. Can you report the opposite?
My understanding is that it's pretty much an unavoidable risk due to the presence of radioactive isotopes in the chip package and the presence of cosmic rays. You're just more likely to actually see it if you admin a thousand machines handling mission-critical data and running 24/7/365 than if you admin one machine running office apps 8-12 hours a day. There are probably millions of users who have had a soft error on their machine and never knew it because it caused some trivial error like making a pixel slightly more blue for 1/60th of a second.
Originally Posted by chithanh
I have had a RAM module going bad on me on my desktop. Luckily, it appears that it immediately caused the system to crash and not boot up again.
Had it instead silently corrupted my data over weeks, the situation could have been much worse.
I once had a board that would corrupt memory if all the slots were loaded and run at full speed. It happened maybe once for every couple dozen gigabytes of data processed, and I don't think Memtest86 was ever able to catch it happening. That was a fun one to diagnose. I assume that was because of some fudging of specs (capacitave load vs. drive strength comes immediately to mind.) rather than external causes, but ECC would have caught it a lot sooner. I think I used that machine for almost a year before suspecting that something was wrong.