Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!furlong.cis.ohio-state.edu!martens From: martens@furlong.cis.ohio-state.edu (Jeff Martens) Newsgroups: comp.sys.amiga Subject: Re: multitasking on wallstreet Keywords: IBM PC RAM parity Message-ID: <38072@tut.cis.ohio-state.edu> Date: 3 Mar 89 20:32:59 GMT References: <1295@csd4.milw.wisc.edu> <503@morgoth.UUCP> <32244@auc.UUCP> Sender: news@tut.cis.ohio-state.edu Reply-To: Jeff Martens Organization: Ohio State University Computer and Information Science Lines: 46 In article <32244@auc.UUCP> rar@auc.UUCP (Rodney Ricks) writes: [deleted stuff from various places below...] :Actually, I think that you may be. Parity checking on the IBM PC has a :MAJOR problem. Say the system finds a parity error. What does it do; how :does it recover? : :Recover, what does recover mean? If a parity error is found, the system :puts up an error message: : :PARITY ERROR : :in large, 40-column text, up at the top left of the screen, AND LOCKS up. :Great for real-time systems on Wall Street, huh? : :Also, from what I've heard, Parity Errors on personal computer RAM chips :are EXTREMELY rare. IBM's "solution" seems to just make the problem worse. : :There are some companies that offer error-checking and correction boards. :Of course, the problem here is that the main system memory will still be :just parity checked, without correction or recovery. : :Of course, what's to stop people from making memory error checking / :recovery boards for the Amiga? In fact, I remember seeing some advertised :awhile back. :Actually, I've heard the following opinion from someone: Since the PC :has so many extra RAM chips for parity checking (12.5% extra), the extra :RAM chips INCREASE the possibility of parity errors, and, since not all :parity errors will be detected (what if 2 bits per [9-bit] byte change? :or any even number of bits?), the extra chips for parity detection aren't :worth it, and may even make the system less reliable. You make a number of good points. And yes, adding one parity bit does increase the chances of a parity error by 12.5%, and, if you get a parity error on an IBM PC there's a 1 in 9 chance it's the parity bit that's wrong. Basically, although I'm sure there are sales types out there who will tell you that you need parity checking, it's not worth doing if you're not gonna do it right, and you can't do it right with just one parity bit. With more than one bit, you can get into correction and stuff, but the cost gets prohibitive for most applications. -=- -- Jeff (martens@cis.ohio-state.edu)