Next time an Itanium machine has problems because of bad non-Intel components, I'll take pot-shots at it too.
I would be surprised if any of the HP IPF server didn't support chipkill memory redundancy. A bad DIMM would be logged and reported but the system would stay up with all processes intact. Same for a POWERx box.
If you want to know why you have problems with your HP hardware, check this: http://www.theinquirer.net/?article=20242 Carly probably thinks it's a good idea to sell servers, tested like a PC (zill testing), yet selling at true server prices. If the customer complains, tell him to buy a “real server“ next time, hoping that the stupid customer will buy an Itanium based HP. Unfortunately, too many customers will follow that advice, rather than buying a true server from Fujitsu-Siemens, IBM or Sun.