[mdlug] ECC RAM failure data - jre

[Jeff Hanson]

On Thu, Feb 26, 2009 at 6:25 AM, john_re <john_re at fastmail.us> wrote:
> Do you use ECC RAM? Do you have any data about failure rates?
>
> I'm evaluating this for a system with 8GB DRAM, &
> http://en.wikipedia.org/wiki/Dynamic_random_access_memory#Errors_and_error_correction
> says
> "Tests[ecc]give widely varying error rates, but about 10-12upset/bit-hr
> is typical, roughly one bit error, per month, per gigabyte of memory.

I had some bad Crucial Ballistix that failed constantly.  It was part
of a bad batch:
http://www.newegg.com/Product/ProductReview.aspx?Item=N82E16820148069

After the last crash corrupted my LUKS/dm-crypt volume I was finally
able to isolate the problem to the memory (after buying another
motherboard).  By that time other people started posting about memory
failures with this batch.  I switched to Kingston KVR800D2E5K2/4G and
have two sets (8GB).  I didn't notice any significant performance loss
but I didn't test for it either.  I set the scrub rate to something in
the middle, 1/day I think.  No idea about the relative error rates but
putting swap on floppies would have been more reliable than the
Ballistix.