On Sun, 2007-01-21 at 12:59 -0400, Clark Morris wrote: > The non-ECC "seems" to be very reliable on both of the currently used > computers at home (1 desktop, 1 laptop).
How do you tell the difference between crashes due to memory failures and crashes due to crapware? Some years ago we had a mixture of servers running NetWare -- some of them had ECC memory and some had parity. The ECC-checked servers never burped, but when the parity-checked servers detected a memory fault they raised a NMI and NetWare stopped hard. This happened on the average of once a month in a population of ten servers -- a demonstrable, no fooling memory check. We also had 300 desktop computers that weren't ECC *or* parity checked. I reasoned that since one of every ten machines failed once a month due to memory problems, then 30 times that many unchecked desktop computers were probably failing 30 times as often -- but silently. So once a day, somewhere out in our small network, someone was getting hosed due to a memory failure. Maybe the machine locked up, maybe it just flipped a data bit and made a mess. Nobody knows. I am sure that our help desk took calls for many of these, but there was NO WAY TO TELL what caused the failure. Usually in such cases the tech would blame it on bad karma or cosmic rays. "Windows is like that", they'd shrug. "Just reboot." Well, Windows *is* like that. But shoddy hardware is a problem too. -- David Andrews A. Duda and Sons, Inc. [EMAIL PROTECTED] ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
